Kerry

That looks like a great idea. I've had concerns about feeding the reference. I've minimized it down to 180uA, but love the idea of removing it and using ground as the reference. I'll model that later. Might want to drop raise the current a bit for the error amp. Thanks for pushing this. Awesome work

Congrats! Really glad to hear that. Looks like this will go onto my list.

Nice read Jim. Thanks.

You should definitely bring the zener string voltage down by about 25V. You're basically removing the value of the pre-regulator and passing a bunch of ripple voltage into the error amp. It will still work, but degrades the performance of the regulator. There tends to be a 10V - 20V variance from cold to warm on the zener strings. If you bring the zener string down to 450 when warm they will likely be around 430V or so cold. You'll need 10V - 20V or so above the 400V regulated out so things are getting very close and I wouldn't recommend raising the regulated voltage. Note that when spec'ing the transformers you should leave about 10% headroom to account for voltage fluctuations in line voltage, so optimally you'd look for 495V or so (also accounts for about 3V-4V of ripple). That's why Kevin suggests 500V unregulated and setting the zener string to 450V. All these extra tolerances are there because actual electronics are different from what we calculate. I've been playing with some ways to lower the drop out voltage a bit. They model well, but complicate things a bit. EDIT: I wouldn't worry too much about your unregulated voltage of 477/479V.

Looks nice. You might want to switch the output connectors to the back side of the board. If I understand it correctly, the inputs and switch face the back side of the amp and the outputs would face inside the amp.

Thanks for the compliments. This has been a really fun project for me. I've been listening to this for the last year or so. I needed to fix the servo and case it up. I finally got to that

I haven't built this yet. I'd guess nothing, 5pF or 220pF. just looking to kill any oscillation if it exits. in the spice models I used 5pF, but that's just a model.

Thanks to the NYC Spring Meet yesterday, I was motivated to finish casing up my KGST. I'm very happy with the result. The chassis is 7" x 12.8". Perfect for my night table Here's the beast in it's intended home: Still need to do the finish and put a respectable knob on it. Here's the back of the amp. Because of the size I wanted to use a single 5 pin mini xlr. Had to build a cable as well. I Put the DAC behind the night table to keep a clean look. I'm just spent the last couple of hours listening and just love it.

Nicely done! How does it perform?

I just built the spice model for the Carbon and noticed the same thing. It should use a 2.5V or max 5V reference.

The net benefit is that it will have a larger operating range.

Just checking the board/schematic around the servo. Looks good. You can change the label for PZTA28 to PZTA06 (same pin out) or some other NPN. Keeping the MJF15030 part is a good idea. Also, the 5K pot should probably go down to a 1K (or 2K) POT and also change the 20K to a 24K (TBD) resistor. Varying by 400 ohms moves the output about 6 - 8 volts. It's just a matter of making it have enough range, but not being too sensitive. The last thing is that I currently have 2pF caps across both legs of the 2 x 1M resistors on the KGST build. I'm not sure if this helps or hurts. I'll try pulling them out this weekend. I'm hoping the 220pF cap can manage any possible oscillation. I'll confirm the rest of the board then built the LTSpice model next.

Good news! I fixed the oscillation on the servo as well. I simply added a 220pF cap across the base/collector of Q3. @pretentiousfood - Looks interesting. I was thinking about doing something against the input stage. Not a lot of current and only about 15V. I'm going to take a look at what you've posted. Thanks.

It seems like tying the reference up the way you did should work fine. I've modeled that approach before. Glad it seems more stable. Regarding the scope, I only wanted you to look at the outputs. You would ground the clip on the probe and measure each leg of the amp. PM me if you'd like and we can walk through the process. I think I have a solution for the oscillation in the KGST but need to try it tonight.

Awesome Do you have a scope you can throw that on? We can tweak some of the values if needed. I'm guessing you were able to recalculate some of the values in the voltage divider chain. I love that you can turn ideas into practice so quickly. EDIT: I'm guessing you added this servo to the third stage (same place the POT was) and not the output stage.

I published the schematic for the KGST w/ servo in the KGST thread here:

I've been playing with the servo in my KGST build. I had a couple of problems. First, at very high volumes, the servo would fail. I fixed that by moving away from the LT-1021-10 which requires about 1.3mA to run properly. I moved over to the LM4040-10 which wants about 150uA. I set it just a bit higher at 170uA. Problem solved Second, I have a small oscillation of about 100mV around 6.7MHz. This one has been a bit trickier for me. I figured I'd post the schematic here. This schematic has the updated cascaded current source, which is currently not on my board. The model runs fine. The schematic uses different values for the sand, but they are closely matched to the actual parts used. Here's where I'm at: The way it works is that I built a rough op amp (similar to the T2 batteries) which functions as an error amp to hold the input to Q1 at 10V (voltage divider via R9, R10, R13, R14 against R11) same as the input at Q2. It uses Q3/Q4 (I've used a single Darlington transistor) to create the proper current (about 20mA) to hold the output at 0V. This is similar in principle to the T2 servo, but need to push/pull in the opposite direction (why I used the error op amp approach). It works great (barring the oscillation) and holds the output within a volt or so from cold to fully warmed up. It is a simple average of the +/- legs. Since the oscillation is common mode, outside the audible range and relatively small, I can't hear it. Any thoughts would be appreciated.

Nice. No reason to be embarrassed Congrats @joamat - I've got to figure out if the servo is possible and stable first. I can post some ideas if you'd like to experiment a bit.

No. It's a very different design and runs standalone. Would need new boards or possibly a daughter card.

The servo I did was for the KGST. I had a couple of issues with it. I just happened to be working on it this weekend. I solved one issue that caused it to shut down at very high volumes. I still have a small oscillation of 100mV around 6.7MHz. It's common mode so I don't think it affects the sound. I'll publish the Ltspice files on the KGST thread later tonight. I have some ideas of how to adopt it for the KGSSHV, but this amp has much less voltage between the bottom rail and the top of the servo.

Some regulators require a load in order to regulate properly. Try adding a 2K or so resistor across the output before you power it up.

It's all about those bass molecules

I might be stating something that you already know, but make sure it is a slow-blow fuse.

It doesn't matter which way you put the pot in.

I've used SOT23 versions for the front end of the KGSSHV and KGST which are lower power than the mpsa series. They get a little warm but run and sound great. I don't see an issue with the mpsa series for the front end of the ES amps.