USB-C experiments – part 2…

As I wrote about a few weeks ago the USB-C power supply standard might open up a few new possibilities for DIY designs. Having recieved some PCBs now, it turns out I was right… 🙂

The first is very much an old idea – a rail-splitter or “virtual ground”. A rail-splitter does exactly what it says on the tin, it splits one supply rail into two equal rails with a “virtual” ground in the middle. Tangent described this idea in a fair amount of detail around 20 years ago and it was used successfully in many of the original Tangent/head-fi designs such as the original Pimeta. Most railsplitter designs have a fair few drawbacks so to me they have also been a bit of an edge-case that I didn’t see much need for.

However, I “reacquainted” myself with the railsplitter concept last year when Neal over at enjon.uk posted about building a “high-power” rail splitter for testing use. I started doing my own version of the layout more or less as soon as I saw Neal’s post because it seemed useful not just for testing, but as it often happens I couldn’t really “crack” making a good layout the first time and so I abandoned it. Picking up the layout again this time worked though, mostly because I started routing with the BUF634 in a DIP package to give me a good basic layout which I could then switch to the SO-8 outline that the new BUF634A comes in.

Having managed to crack the railsplitter layout I started looking at my “back catalogue” to see what else could be made to operate on a single 20V rail. The obvious choice was a Pass B1 because it basically does not require any changes except some optimised power filtering. I might still get to that at some point, but I realised the JFETs for the B1 are even harder to find than they were back in 2013 so a new design would be of limited use.

It then occurred to me that something else that has moved on since I first learned about railsplitters is opamp-design. The original constraint was the the TLE2426 doesn’t really like supplying more than around 20mA, but as most of the newer opamps has a much lower power consumption and so even driving headphones should be doable. I therefore decided to convert my “MiniPre”-design to single supply using the TLE2426 railsplitter. Originally I started looking at the TLE2426 in TO-92 format because it’s very compact, but the noise reduction graphs in the datasheet quickly convinced me that using the “full” 8-pin TLE2426 and the noise reduction capacitor would be a good idea when the source PSU is a potentially noisy switching unit.

My prototype amp uses the OPA1656 and I’m quite pleased with how this one turned out to be honest. I only had to extend the original board by around 6mm (0.25”) in length to fit the new PSU arrangement. While I’ve only briefly tested the amp it works very well, even on USB power. No noise, hum or outrageous turn-on/turn-off thumps while running off the decidedly non-spectacular PSU that comes with my HP work PC :).

This isn’t an “endgame” design by any means, but within the limits of the topology and space constraints I think it’s a very capable amplifier. USB-C supply clearly isn’t an “endgame” PSU solution to begin with and you might get a little bit better performance from these opamps at a higher supply voltage. Very heavy headphone loading is probably also a potential issue, but using this as the basis for a small monitor-controller or pre/headphone amp would be pretty obvious. If you can find a small case and shoehorn it in, I guess it would even work well for occasional office use without having to worry about mains supply and carrying a external PSU.

Project files for both of these designs coming within a week or so once I’ve had a chance to compose my thoughts on them 😀

27 Responses to USB-C experiments – part 2…

  1. Neal says:

    I thought this amp was already a great layout, but it’s really nice to see you keep revisiting and revising. I’ve been sat on a couple of OPA1656 for well over a year now, this might spark me into doing something with them.

    • theslowdiyer says:

      Thanks Neal – and thanks for the inspiration this time round. If you want a couple of my spare boards to get those OPA1656s playing some music then just let me know, because I definitely will not need all of them 🙂

  2. Neal says:

    Thanks, that’s very generous of you, I will take you up on that. It will be good to do a direct comparison.

  3. Pingback: Project files: Single-supply experiments… | theslowdiyer

  4. Sergi says:

    Hi from Barcelona!

    Very interesting topic! I’ve been thinking about avoiding mains for the preamp that I am building, and this really comes in handy.

    How is it different to Neal’s and Tangent approach, that use the buffer besides the opamp as feedback loop? More than the how, why did you decide to remove it? At least that’s what I interpret from the v1.0 picture, I might be wrong though because you mention the BUF634 in the text.

    Thank you so much for your projects, they’re really inspiring.

    • theslowdiyer says:

      Hi Sergi, apologies if it’s a bit confusing in the post but I am not sure I understand your question:
      The rail splitter board includes the buffer, just as in Neals original schematic.
      The preamp board does not include the buffer. You could put it in the railsplitter but that is not needed here – the TLE2426 alone is enough to power a single opamp.
      There are also designs that use the BUF634 as an output stage for an opamp (I have one here on the site) but that is not really possible without making the “MiniPre” board much bigger, so it would sort of defeat the purpose.

      • Sergi says:

        Hi again.

        Sorry I probably wasn’t clear enough. I am only interested in the rail splitter board. Also, maybe I’m looking at the wrong pics.

        In the picture, I only see the TLE2426 rail splitter IC and a TL081CP opamp. If I see the pics from Neal’s board, we can see the TLE2426, a LM741CN (I guess that’s roughly equivalent to the TL081CP opamp) and the BUF634A. Maybe in your board the BUF634 is soldered in the other side?

        Thank you!

        • theslowdiyer says:

          Ahh, yes. The buffer is on the buttom of the board because it gives a cleaner layout. The new BUF634A-version is only available in SMD-packages anyway (I’ve used SOIC which is the easiest to solder by hand).

          • Sergi says:

            Ah OK that explains it then, sorry for the ambiguous initial question 😀

            Can I use this to power more than one board? For instance, I will have 3 boards, Volume control, Phono RIAA and HPA. In case I need one rail splitter for each, should they also require a separate power supply input? Asking because in ESP project 43, Rod says “NOTE: The external power supply used must not be used to power other equipment along with the circuitry attached to this project. While multiple PCBs can be powered from the splitter, they must all use the same supply voltage (e.g. ±6V). Attempting to power circuitry with different supply needs (such as a single +12V supply) as well as the adapter shown here may lead to the power supply being overloaded or short circuited, and all devices powered will likely malfunction and/or be damaged.”

            Thank you again!

            • theslowdiyer says:

              I can’t see why powering more than one board would be a problem, as long as: 1) You stay within the current capacity of the splitter.
              2) All boards receive the same supply voltage from the splitter.
              3) All boards are referenced to the same (virtual) ground of the rail splitter.

              I think the last point might be the more likely to be missed. I have a small preamp-design using the single-supply mini-pre that I can’t get to work because I screwed up the grounding scheme between the preamp and the source (a BT-receiver board) 😦

            • Sergi says:

              (I think I cannot nest more reply levels)

              Thank you again for your help!

              1) Current needs, I still need to do the maths. How much power can your board drive? Does it depend on the buffer or the output cap?

              2) Same supply V. I think that’s fine since all the boards will need ±15VDC

              3) Same virtual ground. I am not sure about that one. Music source will be an input selector (phono, dac, aux) that runs on 5VDC. I don’t intend to use BT.

  5. Sergi says:

    (Input selector will be powered by another PSU, of course, since it is single rail and 5VDC)

    • theslowdiyer says:

      Reply to above 🙂
      1) The board can drive up to around 150mA per rail continuously (limited by the buffer) and peaks of up to app. 250mA. If you draw more than say 100mA or so, i would keep an eye on the temperature of the BUF634A as it may get hot (and then shut down). I haven’t done any tests here, but keep an eye on it and consider mounting the railsplitter upside down to give the buffer more airflow.

      2) Yes.

      3) That should be fine as long as you use a separate 5V supply that has nothing to do with the railsplitter.

      • Sergi says:

        Ah good idea on mounting upside down.
        So, I think I have a project to build 😀
        Thank you again!

      • Sergi says:

        Hi again! Sorry I think it’s the last set of questions… 🙂

        1) is the output regulated or unregulated? asking because the Volume controller that I mentioned earlier accepts either ±15VDC regulated or ±18VDC unregulated (makes sense).

        2) if I wanted to build one of these boards for higher voltage output, i.e. the ±18VDC above, is it possible with same BOM? I expect the ICs to be the same but maybe adjust values for C/R maybe?

        3) does the buffer have a thermal pad to solder to? I was having a look at the datasheet and for higher current boards it recommends it.

        • theslowdiyer says:

          1) Depends on what you feed it. It is regulated if you feed it from a regulated PSU.

          2) You need a 36V-capable opamp, the rest will be the same. Note that the BUF will get even hotter at 18V though.

          3) There is no thermal pad because I’ve never had the skills or the tools to solder them properly and I didn’t expect I’d need it. You can probably make a small heat sink for the buffer with a piece of metal and some thermal paste, which probably helps.

  6. Sergi says:

    Hi again and happy new year!

    I already requested the PCBs. Then I added the PSU to Mouser together with the BOM and apparently they won’t send me the PSU to Europe, unless I am a OEM. I had to cancel the order 😦

    So now I am looking for 30VDC PSUs, but all of them have much higher ripple (~300mV) than the one I was about to buy (120mV). Is this going to be an issue for audio? Can you recommend a 30VDC regulated PSU I can buy in Europe?

    Thank you again!

    • theslowdiyer says:

      Hi Sergi, not sure which PSU you have been looking at and why there are export restrictions on it from Mouser, but consider looking at European sources (Reichelt, TME, RS/farnell if you can access them etc.). Another option is to look at whether your boards can use +/-12V instead (they probably can) because a 24V supply is easier to find than a 30V one 🙂

      • Sergi says:

        Thank you! I had a look earlier at Farnell and the cost was huge in comparison to Mouser/Digikey. But Reichelt/TME are OK, didn’t know them. I am looking at OWA-90E-30 by MeanWell, they both have it. Now I see the ripple is 200mVp-p, is that going to be too bad for audio? I suppose I could go for something cheaper anyway. No, I don’t think I can go +/-12V, at least the volume controller requires +/-15V.

        • theslowdiyer says:

          That MW is probably not the best performing PSU you can get, but I see no harm in trying. The risk is that you get audible noise on the output, but if you can’t hear anything it should be fine.

          • Sergi says:

            I see, lucky that Mouser cancelled my order then 😀
            Could you maybe recommend a decent 30V regulated PSU for audio? I’d greatly appreciate

            • theslowdiyer says:

              Unfortunately there’s nothing that I have hands-on experience with to be able to recommend.

            • Sergi says:

              No prob, I just thought you were also using one of these for the rail splitter. I will get any of the 30VDC and try. Maybe I have to add a filter before the rail splitter, but let’s just see. Thank you again

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: