Saturday, December 5, 2020

multiple voltages on tap

*Begin infomercial voice*

Are you sick of your family not being able to agree on a single voltage?...

Want to convert your single channel power supply into a poor man's multi channel unit?...

Would  you like to save a little time breadboarding when you need multiple voltages?...

... Then build yourself a variable voltage multi-tap!

*End infomercial voice* (unless you really don't want to)

Ok, so what exactly is this? It's a multiple output (4 in this case) linear voltage source that share a common ground among them, intended to be used with a regulated power supply at the input. The output from this multi-tap is really only as regulated as the power supply that it's connected to, other than the ability to reduce transients to some extent as will be described a little later. So how does one know what the output voltages actually are? After all, there's no gauges or fancy displays. Well, that's what your multimeter is for! While I'll likely use this exclusively with a variable bench power supply, this could in theory be used with batteries, "wall warts" or other fixed voltage power supplies. Beware: this has some significant technical limitations so care and caution must be observed if you intend to make one! All responsibility lays on the user! 

Now lets take a peek inside and also take a look at the schematic which is drawn similar to how it's laid out.

This is the classic common collector (voltage follower) circuit, where the voltage between ground and the transistor's emitter terminal roughly equals the voltage applied to the base of the transistor, relative to ground. A variable voltage to be supplied to the base of the transistor is accomplished by using a potentiometer (R1 in drawing) as a voltage divider, where one end of it is connected to ground, and the other end connected to the V+ input voltage jack. A voltage anywhere between those two values can be supplied at the wiper output by turning the potentiometer's knob one way or the other. Again, since this is a voltage follower circuit, the voltage we'll get at the V+out jack will match the voltage at the output of our variable voltage divider. Technically, there will be a a slight drop, depending on the transistor used, which is inherent to the way transistors operate. Therefore, the maximum output at V+out will be roughly 1 volt less than V+ in.

Also attached to the base of the transistor is another resistor (R2) and a capacitor (C1). This creates a circuit known as a "capacitance multiplier", but I think is better seen as being a low-pass filter attached to a voltage buffer. It is this low pass filter which enables the output to filter out input transients as I mentioned earlier. To elaborate on this a bit more, basically the capacitor has one of its terminals connected to ground, while the other is exposed to a voltage above that determined by the output of the voltage divider. However, it takes time to charge/discharge the capacitor, which is in part determined by R2 (known as the RC time constant). Because it takes time to charge/discharge, the voltage between the capacitor's terminals must lag behind any change in voltage above a certain rate supplied in front of the low-pass filter (for simplicity's sake, this is at a point between  R1 wiper and R2). Furthermore, since the capacitor's positive terminal is attached to the base of the transistor, and the transistor's output voltage (emitter in this case) "follows" the base voltage, the output also lags behind any changes in voltage at the transistor's input, effectively reducing input voltage transients above frequencies determined by the values of R2 and C1. In the case of the circuit above, this is roughly 2.5Hz.

It needs to be said that this circuit is very simple, and does little else than supply a voltage at a lower impedance than the voltage divider. There's NO protection circuitry, so it will gladly release magic smoke if you neglect to operate it within its component's limitations (max input voltage, current, and power dissipation  ratings). In other words, this is a "dumb" circuit. Luckily, the circuit is so simple that it's easy to repair when you do fry something.

As for this specific build, you can see there's no PCB, and yet, despite its simplicity, was still somewhat tedious to assemble (unless I was willing to let it look more like a rat's nest). So in the future, any new variants will be on a PCB. The capacitors might look comically large for their value, and that's because these are rated for 400V. Of course, the rest of the circuit is not rated for anywhere close to that, but those were the only caps I had, and I forgot to order something more suitable when I purchased the other components. The transistors are actually NPN darlington pairs (BC517) rated for 30V between collector and emitter, 1A, and 625mW at 25 degrees C, and a minimum DC current gain of 30,000. The components were chosen for input voltages up to around 20V. It's not particularly suitable for a 20V input with significantly lower output voltages unless required currents are similarly low. I should probably plot some graphs and stick it to the box.

In the future, I might want to make a multi-tap that is both a variable source and sink, which could be handy when circuits/instruments use multiple op amps that require different rail voltages or biases. I'd probably also choose some more sophisticated and higher rated power transistors that have fancy features such as built-in short circuit and thermal protection.

Saturday, June 6, 2020

electrets and stupidity

So it dawned on me several weeks back that electrets might be a crucial component of my big project which I'm not ready to detail yet. For those of you that don't know, an electret is a dielectric piece of material with either a permanent electrical charge or one with its dipoles arranged in an order that reinforces its electric field similar to how a permanent magnet does with a magnetic field. Historically, the most common use for electrets was for a type of condenser microphone, where one of the plates is made from this material. Another example is adhesive-free graphic signage films like ClingZ. Now, its biggest, and perhaps most important use is in filtration products, such as the N95 face masks we've become so familiar with as of late.

So as I stated, there's two types of electrets: those that have their dipoles aligned, and those that have additional charges basically embedded in the material. The first is usually made by curing the material between two plates at a high electric potential difference. This aligns the dipoles, and once cured, get locked in place. The other method, in its simplest form, places the dielectric material between a conductive surface at ground potential and an electrode at a high positive or negative voltage depending on the type of charge you want you electret to have. The electrode is chosen so that it's prone to corona discharge, ionizing the air, and accelerating it into the dielectric. A newer method is the triode style, where like its namesake vacuum tube, places a "grid" between the dielectric and the discharge electrode. Presumably there's a whole bunch of variations of this method, but they share the key similarities.

At first, I was looking to purchase some bulk electret material, but all I could find was the ClingZ sheets. Once I realized that making my own was likely quite doable using the corona discharge method, I started leaning towards that, particularly since I could choose not only the type of charge, but also the field strength. What really sealed the deal was that by using a motorized drum as the ground plane, I could potentially create electrets with charge gradients; something that I don't think has ever been done intentionally, but could prove extremely useful in my project.

The next several blog posts will detail the work that is going into the project. Needless to say, it started out very hack-y. It has progressively gotten less so, but some stupid decisions were made early on that got cemented into some of the hardware, which should have been avoided with a bit more forethought. 

Tuesday, May 12, 2020

replacement ion + thermocouple gauge controller

In an older post I showed my original hot filament ionization gauge controller and the abuse it took during shipping of my vacuum chamber setup. Despite the fact that the only likely damage preventing it from achieving operational status was most of the front panel switches and some broken LED leads, I decided to relegate it to spare parts duty, and instead purchased the same model with some different options for $60 shipped.

Despite it looking more modern than my original, it's actually from the same period. This one just had the -1 dark grey finish option specified. The most notable difference with this one is the inclusion of two additional needle gauges to the left. I mentioned in the previous ion gauge post that some of these controllers include circuitry to read thermocouple-type vacuum sensors, and in the case of this unit, it can read two of them, though I only have the need for one. So this unit covers all the bases I need for high vacuum measurements, and the rest of the range can be covered by a mechanical diaphragm gauge. There are a few options that my old unit has that this one doesn't: some process control related stuff. For example, I could set some pressure trigger points that would activate relays for the purpose of automating some tasks. Nothing I particularly care about. Because of that, this unit is missing some relays, but otherwise has all the necessary trim pots. However, the front panel doesn't have the requisite holes to get access to them. Jeez...

Not everything has been rosy with this unit, and I haven't even powered it on yet. When I first removed it from the box, it was further packaged in a few layers of Saran wrap, but I heard something rattling around inside. After peeling away the polyethylene plastic, I found that neither the top nor bottom lids were screwed on. Well, I was going to open it up anyway, so I guess that's.... convenient?

The thing that was rattling around, but had already been corrected before taking these photos was that the single relay on the right side PCB fell off its base somehow despite it having been secured with a tie wrap that was too tight to slip over for reuse (so what you see is a new one). Also, the ribbon cable towards the bottom that connects the two boards popped off. Easy fix, and luckily my old unit had a plastic retaining clip that keeps the connector secured. Lastly, my original unit had a small shield that covered the sensitive electrometer circuitry in the upper right hand corner. Strangely, I don't see any markings/scratches on the PCB around the hole where it should have been attached from any fastener hardware. It might never have had it. Still, I have a shield so I think I'll use it. Other than that, I have the necessary fasteners for both lids, and the missing fuse holder cap.

I did just notice that unlike this unit, my original one didn't have solder masks on the PCBs. So despite the fact that the manual I have very much covers this unit, I'd say that this one is definitely newer.

The one other thing I'm aware of that I need to address is the thermocouple sensor interface which is via the the left hand card edge connector at the rear. I found one on Digi-Key that I think is suitable, but I'll wait to purchase it until I have a few more things to add to the list. Then I'll need to come up with a design for a 3D printed housing to securely attach it to the controller.

Friday, May 8, 2020

feeling overpowered

Finally took a bit of a better look at the 3 channel filament power supply from my vacuum chamber purchase, and I realized that it contained 3 variable autotransformers rather than my initial claim of wirewound potentiometer. In retrospect, it makes a lot more sense.

From what I could find online, SEM filaments tend to operate from roughly between 3 and 4 volts at 5-ish amps. As we can see from the second image, the isolation/step down transformer puts out 24v. I didn't check the circuit routing in the filament holder, but based off the way the two leads were coming off of it, I suspect that the filaments are wired in series, which makes sense in light of the isolation transformer output.

This is not going to be suitable for my needs, so I'll need to figure something else out. I could potentially use a computer PSU and add an additional variable voltage regulating circuit off the 5V output, but I'd also need to convert it so that the outputs are floating instead of being tied to ground as is typical. Since the only spare computer PSU I have available apparently has a bunch of electrolytics spewing their guts, it might not cost much more to buy a decent suitable bench power supply off eBay.

At least I'll have a few center tap 24V transformers, variable autotransformers, and 5A AC panel meters in my parts bin for possible future use.

Sunday, May 3, 2020

mystery flange

I've been disassembling the vacuum chamber system to transport it home more practically. I didn't exactly want to remove the ion gauge, but I was reasonably worried that I would break it if I didn't. After I got home I took a better look at it and saw a mating flange style I wasn't aware of. Several hours worth of google searches didn't reveal anything, and so I still have no idea what kind of flange is on this.

On the gauge is a male-type plug with a small step that has what appears to be a brass-like gasket. It did not easily come off, and I'm not sure I should even try to remove it, therefore I don't know if it has any kind of Conflat style knife edge. On the chamber side is a female socket. It doesn't have a knife edge, but it's possible the the first step is ramped, and it has a pretty sharp edge, but it's hard to tell if that's actually the case. 

Anybody know what kind of flange this is?

Saturday, May 2, 2020

sourcing electrons

In my last post I mentioned that I received two packages in the mail. What was in the other?

Tungsten SEM filaments!

I can find a bunch of references to the Cambridge S4-10 (or S-410), but very little about the unit itself. Ah, well. It's not particularly important. While I'm not building a SEM, my planned experiments will require an electron source, so these filaments are a good option. And since I purchased this unused lot of 20 for $15 shipped, it's economical too. I think this supply will outlast my needs.

quick recap

Two packages arrived at my doorstep today. The first, with Digi-Key logos on the box were the electrolytic capacitors I ordered for my TCP 270 turbomolecular pump power supply.

A veritable mix of Panasonic, Vishay, and United Chemi-Con

While it's probably overkill to to have chosen 105°C parts due to the very well ventilated power supply case, the increased price over ones rated for 85°C is negligible for a one off project like this. More annoying is price and limited selection of axial caps. If I was just trying to get the job done and all I had were radials on hand, then I would have happily stuffed those in. 


Who needs solder mask? That might help prevent traces from peeling off when reworking the board, and we wouldn't want that! The annular rings seemed unnecessarily small, and of course the holes were unplated. There were a few ring to lead spots I had to bridge because they got ruined during component removal.   


I had to avoid setting one of the larger caps that replaced an old maroon part (center right) flush with the board due to a ruined trace on the top, so that I could actually get an iron on it. Honestly, I don't know how they could have had confidence that there would have been an acceptable solder joint between the cap lead and the trace in the original assembly.  

I somewhat uprated the large cap. Originally it was 4700uF, which I would have left as-is, but it was rated -10/+50% anyway, therefore I opted for a +/-20% 6800uF because it was easier to find at a reasonable price. The original cap had 4 leads, whereas the replacement has 2. Two of the holes had adequate spacing (though slightly too small for the leads themselves), but the unpopulated relay socket seen to the right of the cap got in the way. So instead I had to choose two different holes which were spaced slightly too far apart, and the hole on the negative side wasn't actually connected to any trace. So I had to wallow out the holes a bit to get it to fit. Also, a small strip of copper tape and solder connected the negative lead to the negative trace. It was a bit of a pain in the ass to work on because there's wires soldered to the board on all sides, and the only way to increase accessibility would have been to disassemble the unit more than I wanted to.

I slapped it back together. Hopefully there are no unpleasant surprises when I finally get a chance to use it.

Thursday, April 30, 2020

current-carrying noodles

As I await my order from Digi-Key for my turbopump power supply re-cap project, I decided to address the other known issue related to that subsystem of my vacuum chamber: The power supply to pump interface cable. Fortunately, included in the vacuum chamber lot were both cables for the power supply, which are rarely included if you buy a used pump or power supply. Both cables are proprietary, and thus are expensive if you could even find them.

Unfortunately, one of the cables did sustain some damage. There was a cut through some of the outer insulation, as well as the insulation and a few copper stands of some of the wires. It wasn't too bad, and the cut wires were unlikely to cause any shorts, though I couldn't help but to think it would be stupid to take any chances that could just maybe result in expensive pump damage.

The two ends of the cable. I already unthreaded and slid back the cannon plug backshell 

Outer gray insulation jacket already cut back. Nicks in individual wires visible. 

Since the damage was closest to the circular connector, that's the side I'd have to fix. I deal with this style of connector quite a bit at work, and before I removed the backshell, I was somewhat worried that the individual sockets in the connector housing would be crimped to the wires, thus requiring replacement. Not a big deal if it has a US military part number, but this thing was made in France.... It ended up being soldered, and once I knew better, I wished they actually were crimped.

I ended up having a few issues, partly because I'm currently so ill-equipped at home. No vise. No solder sucker/vac. Those would have come in handy. I tried wicking some of the solder from the cups using wire. It sort of worked, but not ideal. The wire gauge is pretty close to the limit of what the cups can accept, which make it more annoying to work with. Secondly, I didn't have much wire length to work with after cutting them down and stripping them because I wanted the backshell strain relief to clamp down on the outer insulation. Once I had a few wires soldered in place, it pre-loaded  the other wires making it somewhat of a pain to work with. The only saving grace is that I only had to deal with the sockets around the perimeter of the connector, as the inside ones are unused. I definitely wasn't ecstatic over the results, but they were acceptable.

Finally got it all soldered up. Wait a second.... When the hell did I remove the backshell? 

I decided it would be easier to de/re-solder the connector on the other end so that I could slide the backshell on.

Alas! All done. The wires ring out good with low resistance, and no inter-wire shorts. Now if only I could remember what the hell I did with the little cap that goes on the backshell.

Monday, April 27, 2020

power supply to the people

As mentioned before, one of the critical bits missing from my vacuum chamber purchase was a power supply/controller for the TPU 110 turbomolecular pump. Having looked through the pump's manual, I found that it called for a Pfeiffer TCP 270. Once again, eBay was my first and last stop. Perhaps somewhat surprising, there were quite a few available. I know this particular model covered more than one model of pump, so maybe that explains the selection available. The strange thing I've noticed as of late is just how wide of a spread in pricing exists when searching for specific items. One seller might be asking $30, while another is asking $1000. Sometimes the more expensive items are unused or rebuilt, but typically that's not the case. Nor is it necessarily the case that the more expensive items are in better condition. Often even those are listed as untested and sold as-is. Perhaps those sellers "know what they got"; in other words, something likely priced too high to sell.

Once again I opted to go the frugal route, and found a power supply that appeared to be in decent condition, with a "or best offer" option. Keeping with my purchase price modus operandi, I offered an amount that would place it slightly below $100 after shipping. My offer was accepted and I soon received it at my doorstep in a well packaged state.

Huh. The corner was bent, the gauge needle sat almost midway up the scale, and the "pumping unit" button was stuck pushed in. Who knows when the damage was sustained, but given the way it was packaged, it wasn't during shipping. At any rate, I didn't care about the corner or the gauge, but obviously something would have to be done about the switch. I popped it open to take a look (actually, I did that before I even noticed the switch)

Pretty good cable management, and nice old-school wire lacing 

Yep.. The switch took enough of a beating to crack the plastic nut. Removing the nut revealed access to a pair of retaining clips for the switch face. This is how custom switch faces are installed, as well as how the indicator lamp is replaced. I found that the switch operated just fine without the face installed, but there was nothing visibly wrong with the face either. After reinstalling it and toggling the switch numerous times, it seemed to work ok, but every once in a while it would bind slightly, or it wouldn't latch when trying to engage (it's a push-on push-off style). It was usable, but I decided to swap the two switched, especially considering my pump didn't have the heating jacket option anyways, so that switch would go unused.

This was a great opportunity to try out my new soldering iron...

Behold the power of bacon! Er, Bakon. Bakon 950D to be exact. Several years ago I donated my soldering station to the local makerspace, which was fine because I did all my soldering there anyways. Having decided to build up my home lab again, I was in search of an iron. I fell in love with the pencil design of the Pace ADS200. I came so close to pulling the trigger on buying one, but ultimately I decided against it. Realistically, at this point in time, $300 could buy me a decent iron and a bunch of other things. So then I started gravitate towards the TS100 or TS80, but despite the favorable reviews, I really didn't like that grounding the iron was kind of inconvenient. Next up was one of the many variants of the KSGER with a T12 type tip. But there were a bunch of different ones, made by who knows how many different manufacturers, with different power supplies, using different microcontrollers for temp control, and none inspiring too much confidence. Then somehow I came across the Bakon 950D (listed as Feita 950D on Amazon). While I'd probably prefer a unit that uses T12 tips, this one uses T13. Reviews were good, power supply design was proper with name brand active components (the caps not so great), and actual tip grounding. Per this thread, the unit drives the tip with a power of ~42 watts. The original poster of that thread also modified his, bumping up the output power to roughly 66 watts. Ultimately, it was his opinions of the iron that made me choose it. I might eventually replace the included pencil with something that uses T12 tips, and perhaps an almost ADS200-like finger to iron tip distance. 

So the unit might not look very impressive, with it's simple 7-segment displays, two push buttons, and it's wannabe laptop charger enclosure, but I was immediately pleased it during my first go. I've never used an iron with a directly heated tip, so that was game changing for me. Not only does it heat up significantly faster than anything I've ever used before, but it transfers heat to solder joints incredibly quick too. So here's the kicker: I bought the iron, which included 6 different tips, for a penny under $35. Incredible. I'd definitely recommend it to anyone in need of an iron, but doesn't have the budget for a Pace, JBC, or a higher-end Hakko.

Back to the pump power supply. Scrolling through the pump's manual, I came across this little bit:




Well two can play that game!

Problem solved.

Maybe not that easy, but close! This looks like a job for........

The ol' universal hacking device chip.

Dropped right into place. No soldering required. 

But all that solderless work was for nothing, because.....

Well that was easy! I was actually aware of this before ordering the power supply, because most of the ones for sale were set for 1000Hz, with the few 716Hz units tending to be more expensive. Even before learning about that resistor, I figured the difference between the two would be minor. Luckily it was much simpler than I originally assumed. 

TODO: change the electrolytics -- Just in case. Clean the dip sockets and IC legs. They look slightly corroded. 

Sunday, April 26, 2020

getting pumped

I knew it would be prudent to take a better look at the Pfeiffer TPU 110 turbomolecular pump that came as part of my vacuum chamber purchase, especially given the issues sustained during shipping. Presumably for the sake of providing clearance to the rack mounted electronics, there was a sort of Z-shaped duct to push the pump further towards the rear of the cabinet as can be seen here:

That meant I simply couldn't look down at it from the inside of the chamber, so I had to remove it. After removing it, I found that there was a conflat copper gasket still attached to the flange on the pump, and another on the chamber side flange. I'm not aware of stacked gaskets being acceptable practice. There was also a groove cut on the upper side of the pump's gasket, indicating that it had been reused.

After removing it, I found that there was a splinter shield at the pump inlet, in conjunction with the coarse shield at the base of the chamber. The knife edge on the flange appears good.

Happily, the splinter shield was clean of any kind of crud or debris, and in excellent condition. Though more importantly, I wanted to verify that nothing too obvious was wrong with the pump itself. Thus, I removed the shield and gave the pump rotor a twist. It spun with buttery smoothness with not a hint of grinding. Granted, things that are imperceptible at an insignificant RPM, might manifest themselves in a more dramatic manner when this thing is running at full tilt under load conditions, maybe somewhere in the neighborhood of 40,000 RPM.

I did have a slight concern about the bit of oil found at the outlet port. The oiling system consists of two bearing chambers on either side of the motor, which have some sort of wicking material to provide the bearings with lubrication. Based on the internal profile drawing found in the manual, there only appeared to be a sort of splash shield separating the bearing chamber and the rotor/stator section of the pump, which seemed slightly bizarre to me considering that it is acceptable to mount the pump on it's side. I don't think anything is actually wrong, but I can't say for sure.

All that said, first impressions are overall quite positive. I have reasonable confidence that the pump is in good, operatable condition. However, I will at least give it an oil change.

Friday, April 24, 2020

department of trustworthy marketing

I've been wondering if package delivery people have been getting tired of delivering to my address as of late. Maybe I should leave some milk and cookies out on a table next to the fireplace as a sacrificial offering. Perhaps next time. For now they'll just have to be satisfied with my autograph. That's actually a lie; they're not requiring acceptance signatures due to COVID-19.

Obviously I mention this because I must have received yet another thing in the mail. Another eBay buy, naturally. Nothing too fancy, but rather just fancy enough.

What we have here is a GW Instek GDS-1022, a basic, somewhat dated and low-end 25MHz Digital Storage Oscilloscope. For just shy of $100 shipped (not including $15 worth of probes), I'd argue it's a decent deal, and ticks pretty much all the boxes to suit my needs. As it turns out, these $100 purchases are somewhat of a reoccurring theme with me. You can get an impressive amount of hardware for that money. One thing I forgot to mention in that whole vacuum chamber debacle is that I received a partial refund from the seller due to the damage caused by lousy shipping prep. How much did I get back from the $260 winning bid? $100, of course.

Back to the scope. It's about as basic of a DSO as they come. It's a 2 channel, with a 25MHz bandwidth, and a claimed 250MHz sampling rate, with an abysmal 4k points recording length. The display is 5.6" with a fairly crap 320x234 resolution, which isn't surprising given that this thing is probably from the late 2000s. I didn't merely choose this device only because it was among the cheapest DSOs I could find (it was), but also because GW Instek from Taiwan is arguably a bit more reputable of a brand than some of the cheapies hailing from mainland China and sold new via eBay or Amazon which have pretty much non-existing warranties. Better to be out $100, than closer to $300.

Like far too many other software-heavy devices, the GDS-1022 shares most of it's technical specs with it's "higher-end" brethren. Meaning that this, and every other model below the top range 100MHz unit may have it's bandwidth limit set artificially via software. I can't say for certain that this is the case, but I think there's a very good chance that it is.

Things that I like:
Overall the menus are simple and intuitive. Separate volts per division and vertical offset knobs for each channel is appreciated. The unit isn't some heavy beast, but I think that thanks to it's physical depth, its quite stable when pushing buttons. The same can't be said for some other modern scopes. No frills. It does basic oscilloscope stuff but with the digital niceties such as various signal measurement capabilities and transient capture+store not found in analog units.

Now that I know it actually works, it's time for a quick teardown.

Here's the back cover removed. All it did was add some rubber feet, stickers, and a consistent look to match the front plastic. Seemed a bit unnecessary, but a nice touch that maybe adds a little bump protection. The top, bottom, and sides are all exposed chassis sheet metal.

The top/sides of the chassis is held in with 3 screws and, with a little effort, slides right off. I like the sheetmetal design and quality, and find the overall mechanical construction of the scope to be pretty good. As for the electronics, the rear side unsurprisingly contains the power supply (including LCD backlight inverter) save for a USB and calibration coax breakout board, with the rest of the electronics mounted to the front.

I'm not an expert, but the power supply seems like a pretty standard fare, with nothing too egregious going on. Sure, there might be a TO-220 or two "flapping in the breeze", but Dave Jones doesn't need to know about it. All the wires are terminated with a connector, which tends to be appreciated. The electrolytics appear to be Chemi-Con brand. Sadly, the same can't be said for the main board.

And here's the big money bits. On the digital side is an Altera Cyclone II FPGA and a Max II CPLD, plus some flash and ram. No idea what the CPLD is for. Interestingly, there is no CPU IC anywhere as far as I can tell (unless it's hiding on the other side - doubtful), so it likely has a soft core implemented in the FPGA. The analog input section is probably a partially discrete design? Most of the ICs are pretty standard op-amps save for a 4:1 buffered "video" mux. Immediately to the right of the shielding and doing some hocus-pocus to the input section are 8-channel analog mux/demux's.

Then there's the (single) ADC... It's an Analog Devices AD9288-100; a 2 channel 100MSPS part. But wait, GW Instek claims the scope has a 250MSPS sampling rate. What's going on here? They're either overclocking the chip by 250% (doubtful), or they're overclocking by 125% (a little less doubtful), then, by the magic of marketing, adding the sampling rate of the two channels together and claiming 250MSPS. The manual, unsurprisingly, does not specify. Also, from just my quick glance at the PCB, it looks like the output of each analog front-end connected directly to one of the two inputs of the ADC, so unless there was something else going on in there, I don't think they fed any channel to both inputs, and then sampled each ADC 180o out of phase from each other. It's possible, but the manual doesn't specify different sampling rates depending on how many channels are operating, unlike some other manufacturers. But lets give them the undeserved benefit of the doubt. Still, 250MSPS is pretty much only good enough for 25MHz bandwidth. The 100MHz unit with the same sampling rate would do what to a complex 100MHz signal?

Prior to purchasing this scope, I did some due diligence and tried to find whatever info I could. There was so little info it almost seemed as if they didn't exist. There was a post (I believe on the EEVblog forum) by someone who noticed this:

A 0Ω resistor next to a number presumably representing the unit's model number. In the image above, you can see it next to 1022, my scope being a GDS-1022, while the 100MHz unit is the GDS-1102. As I recall, the poster had a 1042 and attempted to solder the resistor into the 1102 slot. No change was observed, as the scope still indicated it was a -1042. Someone else mentioned that the firmware might need to have been reprogrammed to have the changes take effect. I kept this in mind when buying the unit as that would have been a cheap upgrade. But knowing what I know now, it's doubtful the "100MHz" version would be worth a shit. Still, I'm satisfied with what I have, and I think it'll serve me just fine. Time will tell.