Try and try again…

So after my dismal failure with a cheap TPA3255 board some time ago I thought it was about time to try again. I therefore bought a new TPA-board from Aliexpress – slightly different type than last time but same basic idea – and the price was only around 25 USD.

I’ve just briefly tested the amp for now, but it seems well-behaved with no turn-on and turn-off thumps. This may seem like a minor concern, but I know from past experience that if an amp has major issues here I will never really be happy using it on a daily basis and so it ends up being a big problem.

I’ve also (only) done a quick sound testing with some test speakers. The TPA3255 has a reputation for good sound and this board does not seem to be an exception. Of course a serious audiophile will want to experiment and may be able to find something better, but to be honest for any “normal” person (i.e. non-audiophile) this is could well be “all the amplifier you really need”. Not entirely unexpected, but still pretty impressive for something that is so cheap.

I have plans to case this board at some point and/or buy a second one for 4-channel use, but it may take some time (I am super busy at work right now). One modification I will have to attempt is to remove the tantalum input capacitors (argh!) and replace them with something better – hopefully I can do this with a better result than last time, but wish me luck 🙂

No more class D?

I do most of my electronics-related shopping on international websites, but it still pays to keep an eye on the local classifieds. The audio market in Denmark is small (even though we have a reputation as a big audio nation…), which means that some things are very expensive but also that the occasional deal comes up 🙂

Last week I managed to pick up a batch of NOS Toshiba 2SA1943/2SC5200 power transistors. There are just under 100 pcs of each type and the seller has already done some basic hfe-matching – nice!
They were bought from someone within the industry and I am assured they come from a reputable source (digi-key) so although I couldn’t pick them up in person I felt it was a chance worth taking.

Having now looked at various pictures online they look as genuine as I have ever seen them, so that is obviously great. Do I need 100 pieces of these for my projects? No I don’t, but the price was much too good to pass up so there was no reason to split up the batch 😀

As most will know these transistors are a staple of class A and AB amplifiers where they are used as output devices. And no, that doesn’t mean that I will stop buying/building class D amps as the title alludes to. Class D still has a place and many advantages over conventional linear amplifiers in my view. It does mean lowering the barriers for building some class A and AB amps though, so think of it as an excuse to do a bit more experimentation here in the future 😉

Second chances…

I’ve posted about this before, but much of what you see on the internet is the product of extreme selection bias. I am not less guilty of this than many others since I mostly post stuff after it is tested and confirmed working, but let’s take a peek under the covers for once instead…

On the surface this is my finished amplifier with the ICEedge-based ICEpower 300A2 module. However, what the pictures do not show is that it is isn’t really the same amplifier board I originally bought. I finished the wiring quite late on a weeknight and then violated one of my own rules by testing it straight away. All I got from that was a small puff of smoke and the smell of burning electronics filling the room…

I spent a few minutes looking at my wiring and considered whether to completely abandon the project but then went online and ordered a replacement module instead. After that I decided it was a good time to go to bed….

I normally say that the “immediate feedback” about whether something works or not is part of what I enjoy about all my hobbies and DIY audio in particular. I still maintain that, but had you confronted me with right after I finished a day by letting the smoke out of a 150 EUR amplifier module, I probably would have told you to get lost 😀

Immediately afterwards I thought I had “simply” wired the floating driver voltage the wrong way round. However, after I received the replacement module a few days ago and looked at the diagram and datasheet with fresh eyes I realised I had misread the pinout of the AUX supply connector and badly screwed up the wiring for all the AUX-supplies. Not sure why, but somehow it softens the blow a bit….

Anyway, this “version 2.0” seems to work just fine and I’m in the process of comparing it to the 700AS2. I am also happy to report that the 300A2 case is now standing atop a small (but growing!) pile of stuff that is actually finished and working – more on that later!

Postscript 26/2-2022: I have now done quite a few tests on the amplifier and on the plus side – it sounds really good. On the minus side – when running it single-ended (= in stereo) like this it suffers from what I believe is “bus pumping”. This means that the power supply occasionally shuts down due to overvoltage, which is really annoying to say the least. There is no mention of this problem in the ICEpower datasheet, probably because the ICEpower supply is protected against this problem and the cheap switching supply used here is not.

I’ve done the “standard” trick of reversing the connections to one channel and that helps, but it is still not enough. Next step is to try and add some more capacitance, but not so much that the power supply complains obviously. In short – this is a worthwhile build, but pay attention to the PSU arrangement if you do decide to build it.

This is an ex-amplifier…

A pile of amps…

Not quite a New Year’s resolution, but still something that I have promised myself to do something about this year is my pile of unfinished amplifier projects.

Near the top of the pile – mostly because it didn’t need that much work – was my Pascal power amp (based on the S-Pro2 module). Late last year I got the front panel from Schaeffer which was really the only missing piece and all that was then left was to cable everything up. So to (hopefully) set the tone for the year I got stuck in and finished that this week.

No real complications with finishing the amp, apart from taking 3 attempts to dim the brightness of the ridiculously bright white LED in the switch to an acceptable level. I like the clean look of the white LEDs, but making a standard red or green LED into something that doesn’t burn your retinas out is just so much easier…

Other “lessons learned” include that I was happy I took the time to design a break-out board for the module, because the last bit of wiring was so much easier this way instead of having to air-wire everything to the same 26 pin connector on the Pascal module. Also, I should have moved the module forward another 10mm or so as the space for the mains wiring is a bit (too) tight for my liking, but it should be OK as it is now.

I knew the amp was working from previous testing, but I hadn’t really listened to it. Well, I have now (briefly), and while I still use the ICEpower700AS2 as my main amplifier and I think it has better synergy with my current speakers, there is no doubt that the Pascal module is a very capable amplifier. I no longer have the Hypex Ncore-amp I built to compare it with, but where I rate the amplifiers one of the things I notice is that – like the Hypex – I find the bass a little too pronounced with the Pascal amp. This is definitely a personal preference/synergy thing and the speakers and room will emphasize the bass though.

I’ll probably keep listening for a little while though, but as I was on a roll with building I managed to do a “facelift” of the 700AS2 as well (new front panel and proper switches for mute and standby). The wiring issn’t as pretty as I’d hoped for because I had to wire the LEDs to the respective PSUs instead of to the LED drivers, but once again the breakout board really helps. No changes in terms of sound obviously, but at least now the looks mean I’ll be happy to keep it on the shelf as well.

So, one (and a half) project done and the motivation to continue chipping away at the pile of amps has increased that much more – let’s see how the year moves on from here 🙂

If it’s not broken…

… modify it until it is!

Not sure I’ve ever posted an example of me destroying something that otherwise worked perfectly well. That’s not because it didn’t happen before (on a number of occasions…) but because I think this one is worth showing off 😀

The background is that I bought a small TPA3255-based amplifier PCB along with one of my recent Aliexpress purchases. This is called the Shui Yuan amp (diyaudio-thread here) and it costs somewhere around USD 30 incl. shipping. The TPA3255 has got a reputation as a good amplifier and since I have a few suitable PSUs I put one in the shopping basket.

I received the board, checked it and confirmed that it worked well. Very importantly there were no nasty turn-on/turn-off thumps and the sound quality was definitely promising. Having thus confirmed that it worked and sounded very well, I set about making it sound even better… The obvious (and intended) modding object is of course the input opamps (NE5532s – as usual of questionable origin). These are socketed and so very easy to replace.

However, there are also some interstage coupling capacitors which are Nichicon FW polar electrolytics. They look genuine, but since I have a bag of Elna Silmics that are supposed to sound better I wanted to change those as well. Also the two main power caps were a bit tall for my liking, so I wanted to replace those with some shorter ones to ensure a good fit in a 1U case.

In principle this is all a simple job, but in the middle of the desoldering process I realized that I was not only removing solder but also quite a few of the pads on the underside of the board. Pretty annoying, and very difficult to do anything about on a double-sided black PCB where you can’t really see what you are doing. The culprit here – apart from my lack of desoldering skills – is obviously that I forgot that this is a cheap product on cheap PCB material. It’s not impossible to fix, but it’s not easy and right now I’d probably not spend too much time on it – oh well.

From what little I heard this is a great amplifier PCB at a reasonable price though so if you want to experiment I would definitely recommend buying one (but consider limiting your mods to replacing the opamps … 🙂 )

• 

  This is an ex-amplifier…

• 

• 

Closer to the edge…

As you probably saw I added an ICEedge-based ICEpower300A2-module to my collection not so long ago. I originally envisaged this as a long-tem project because the module requires a fairly complex PSU (with five supply rails) and I didn’t want to “splurge” on the matching ICEpower PSU (too large and too expensive).

However, for a different project I was looking through Aliexpress for cheap switching Power Supplies and I then spotted the option of customising some of these PSUs with the specific voltages you want. That was definitely an interesting turn of events, not only because the Aliexpress PSUs are massively cheaper than the ICEpower PSU, but also because it is much smaller and allows for a 1U case to be used. That gave me a case design idea as well, and so I bought a PSU without having ever held one.

Having received it now it seems to be well-made and the output voltages are more or less spot-on what I requested – great!. The 250-300W power rating is probably a little on the low side, but for a bit of evaluation and home use I think it will do just fine. I haven’t tested how noisy it is, but again for a quick test it’s probably more than plenty.

Now am sure you hear about parts shortages all the time these days and this project is also somewhat affected. It turns out “the missing pieces” of the puzzle in this case are rather small, namely the double-row JST connectors for the ICEpower modules control and low-voltage power connections – very annoying. I couldn’t find them in stock fro any of the “normal” sources that I am actually able to order from. I think I found a place to order these now, but probably with at least two-week lead time and possibly more. Let’s hope they actually show up 🙂

Lots of other things going on the mean time though, including a new PCB project I am quite excited about so even if we are heading into the holiday period then expect more updates coming later 🙂

Full house?

After this short break the shopping channel continues… 🙂

As I’ve mentioned before I’ve owned and listened to nearly all the standard ICEpower modules ever made. The only ones I’m missing are the 300AS1 (the last – and according to some people the best – ICEpower module that’s made purely with discrete components), the TI TPA325x-based 100AS2/200AS2 and the ICEedge-based 1200AS2 and 300A2/400A2. Well, this week I took a step closer towards the goal of trying them all because I received both a 200AS2 and a 300A2 in the mail.

These two modules are of course each interesting in their own way, but the all-in-one design of the 200AS2 makes it a bit easier to work with. As a result, I’ve already got a pretty clear plan for what to do with the 200AS2, and while I – sorta, kinda – also have a plan for the 300A2 it is going to be a bit longer. For now I’ve only done a quick check to confirm that the modules are OK though.

The main reason for the delay in building with the 300A2 is that I didn’t buy the matching 1500S PSU because it was far too expensive. While I already have a main PSU that meets the required specs, the rest of the supply arrangements will need to be worked out. That possibly requires me to do a new board layout, so while I’m anxious to try out the amp and hear what ICEedge has to offer, it might well take some time before I get to that.

I actually also ordered a pair of 300AS1s in the same week as these, but it turned out the seller wasn’t able to deliver anyway and so I ended up with a refund instead. Oh well, the hunt for the last few ones continues then 🙂

Project files: Pascal I/O adapter

After a longer-than-expected delay in shipping, I received some boards. Among them was the revised version of my Pascal adapter board, so after a quick check that it still works, time to release it into the wild… 🙂

Update 18th July 2022: A reader was kind enough to point out that there are actually two versions of the “universal” Pascal 26 pin I/O interface. There is one for the L-Pro (that I was looking at) and one for the S-Pro (which I didn’t have access to previously)

Basically the RX programming resistor is not used on the S-Pro modules and instead pin 14 can be used to put the module into a lower voltage mode suitable for running the S-Pro module in BTL with a 4-ohm load.

The download file has now been updated so there are now two versions of the adapter-board with some minor tweaks for each module type. If you have been using the “old” version of the board with the S-Pro modules then don’t worry. If you want to be sure then remove the RX resistor by cutting it off the board, but if your amp already works then it is unlikely to make a big difference.

I’ve also included some additional information in the file that I did not have access to before so you can see what the differences are (I wouldn’t normally, but this is pretty annoying so I’ll make an exception…).

Update 30th August 2022: Another reader sent me the pinout for the T-Pro modules and we’re now up to three versions of the same “standard” interface…

The T-Pro and S-Pro interfaces are very close, except that pins 14 and 15 on the T-Pro are reserved for internal use. If you have boards already, don’t mount or cut off the RX resistor. If you do not have boards already, use the S-Pro version and don’t connect anything to pins 14 and 15.

Update 13th November 2022: So, the saga of the Pascal-modules and the documentation continues. A reader pointed out to me that the input of the Pascal-modules isn’t actually a balanced interface, but merely has a split of different grounds broken out onto the connector. After reading the datasheets very carefully again, this seems to be correct and I just missed that.

The adapter boards below are still usable, but basically the modules have two SE-inputs and you’ll not be able to run a true balanced signal from a preamp unless you a) use the modules in BTL-mode or b) add a way to convert between BAL and SE in front of these modules. In this regard, the configuration is exactly the same as the ICEpower ASX-series. See further comments below.

ICEmatch-modules…

It’s getting harder and harder to find ICEpower modules I haven’t yet tried, but here is one – the “ICEmatch” amplifier and PSU combo.

Developed as a modular solution for multi-channel and custom-install applications, each 80AM2 amplifier board can be two-channel stereo or BTL mono, while the separate PSU can power up to eight amp boards (= 16 channels).

I grabbed the set (one PSU and two amps) from a local classifieds listing because they happen to match a project I am planning (or should I say “contemplating”) – a sort of large-ish bluetooth-speaker kind of thing for a bedroom or an office (although probably without bluetooth – don’t need that!).

They USP of the ICEmatch boards compared to the normal modules is that the two amp boards I got can be configured into a 2.1 amplifier with a bridged channel for a “subwoofer”, which is exactly what I want here. Also, the PSU is well-equipped with aux. voltages for control and input boards which I expect to need.

This is definitely a long-term project because not only do I need to figure out a front-end for the amplifier, meaning an active filter circuit of some sort, but I also expect I’ll need to lay out a couple of adapter boards for the wiring and work on a lot of other things (the acoustics, for one).

I am quite confident the modules will be good enough for this project and sound quality isn’t the ultimate priority as such, but I might have to rig up a test set up and try them out first nonetheless… 😀

Powering up the Pascal-amp…

I am always amazed by how a project can seem impossible to get started on for a very long time, and then suddenly one day the solution is there and the work of realising it is only a few hours.

One such example was when I bought a Pascal S-Pro2 amplifier module last year. As mentioned in the original post I wanted to create a breakout adapter board for the Pascal’s 26-pin IDC-connector which isn’t the most user-friendly connector around. I’ve actually opened the board layout on several occasions but I really couldn’t crack how to get a good start on it.

In this case it was simply working on a different PCB design (which I expect to show shortly) that suddenly provided the flash of inspiration I needed. The realisation was that I could put the connector for the Pascal board in the middle of the breakout board and then arrange the breakout connectors pins all around it – not exactly rocket science, but there you go.

I’ve just done a brief test on the adapter board and it seems to work well – it is certainly easier to use than the IDC. I will of course be publishing the boards files for this one but it’s going to be some time because it does need a little rework first.

Problem number one is that I’d sort of forgotten that a double-row connector isn’t fully symmetrical so you can’t put it on the bottom if it is placed on top in the layout. Problem number two is that the connector really should be a polarised box-header to match the IDC connector and there’s barely enough space for that. However, the current incarnation of the board seems to work and that at least means I can listen to the Pascal module in raw format and also start thinking about a case layout for it – let’s hope the inspiration for that one comes a bit sooner 😀