KG Balanced Dynahi build discussion thread

[JoaMat]

Added a NPN, a trimmer and some resistor. With those it should be possible to set current through output transistors. Well, that’s the idea. If it works I probably let you know…

[MLA]

What could possibly go wrong ? 

Seriously; great idea though! Things could potentially get toasty on a 2U heatsink, so being able to choose current would be welcome flexibility.    

[JoaMat]

Just want to add that the bias control is not my invention. You find similar on Uberamp and AMB’s beta22.

My view on Copyright is - my right, as DIY:er, to “copy and paste” as I wish.

 

Edited March 23, 2022 by JoaMat

[Bespav]

On 3/21/2022 at 1:42 PM, JoaMat said:

So here is my first attempt

Hi, JoaMat!

Very nice PCB routing concept.

Something similar:

Let's go ahead, it's time to provide DIYers with the next features:

1. A simple swapping JFET/BJT input stage on the mezzonine with standartized fitting place/connector.
2. Really smaller dimensions with easier through-pcb-screwing.
3. Integrated onboard fitting connector for corresponding voltage regulator.
4. Check transistors availability and, of course, create different versions of the output stage, one for TO-126, other for TO-220 devices and maybe one for TO-3P/TO-247/TO-264.
5. Since you have provided bias control it's time to make output stage 2-stage emitter follower, vertical mosfet source follower or lateral mosfet source follower for those who want to experiment.
6. Input JFET transistors have cascodes, so aren't we ready to provide cascodes for VAS devices?
7. VAS devices on the main heatsink and also an option of the TO-126/220 cases.
8. Really higher dedicated IPS/VAS supply rails, up to +-150 volts.
9. An option to use modern low-ESR solid polymer capacitors for the power stage.
10. Dedicated input stage before volume regulator for common mode error rejection.
11. Second order integration filter for output nulling.

 

So, the concept.

Mezzonine construction. One PCB for JFET/BJT input stage and corresponding cascode devices, second PCB for one OPS shoulder and third PCB for the voltage regulator. All for as-LEGO-construction as possible.

Edited March 24, 2022 by Bespav

[JoaMat]

Interesting ideas, wish list, for future Dynahi, Bespav.

Here is a new amp board. Initial tests have been successful. The bias control seems to work.

Heat sink is Modushop 80mm x 200mm. 

[mdr30]

Joamat's build reminded me of some problems with balance and setoff on my own Dynahi. I adjusted the amp a couple a months ago, a bit tricky with DC fluctuating, but got numbers down to 3-15 mV with servo in, depending on volume pot position. However, with headphones connected, DC between hot and cold is 50-80 mV. Anyone got theories why this is?

[JoaMat]

Here is picture of the amplifier in work. Mono version. AMB sigma 22 PSU.

It has four trimmers. Lower middle controls the offset. Middle upper controls the balance. The two outer controls respectively channels bias. Input stage current source/sink approximately 1.7 mA. Output bias 35 mA.

Setup procedure I used:
•    Let heats sink get warmed up to working temperature
•    Trim offset to zero
•    Trim balance to zero
•    Set desired bias of both channels to desired bias

Redo previous procedures if needed.

I find the setup easily done. 

[JoaMat]

...and now a stereo version.

Bias 75mA, 125 degrees F at heat sinks and AMB sigma PSU is working hard.

Might need better chassis.

Edited April 19, 2022 by JoaMat

[JoaMat]

Milled a front panel, with inscription, today. With black painted aluminum net as top cover you can see the light from eight red LEDs.

Edited April 21, 2022 by JoaMat

[bbest]

Hello!

I could get now DynaMight 4pcs boards version 2004. Does anybody know is there any difference for balance amplifier between: DynaMight and Susy Dynahi v 1.43?

Edited May 27, 2022 by bbest

[JoaMat]

A slightly reworked “Dynahi smd” board. Balance and offset trimmers removed. Now servo on each side that bring the offsets to ground potential.  Bias trimmers moved to north for easier accessibility “in flight”.
 

This is a "solder and play" board. No set up procedures required, except to turn bias trimmers to desired bias. At least that’s the idea.

[gepardcv]

Is that a stereo board? I'm trying to wrap my head around how to heatsink it properly.

[MLA]

You have to think Joachim style, i.e., smd-small and vertical . I believe the board is stereo and it mounts flat (vertical if seen from above) to a 2U heatsink. In the pic four posts above you can see how the previous edition boards are mounted through the grille.   

Edited October 21, 2022 by MLA

[JoaMat]

It’s a balanced board.

Inputs in center. Outputs at far right and far left. Positive rail at top edge and negative rail at bottom.

 

Dynahi boards mounted to 80 mm x 300 mm heat sinks. Here I’ve removed upper screws holding front and back plates to heat sinks and then “unfolded” for maintenance. To make it possible to unfold the wires are made “to long” when folded and then it looks like a snake nest. This is an Only DIY version, not to be sold...

 I hope this makes it a bit clearer.

[penmarker]

Received my j74/k170 from punkydawgs the other day. He even included some metal cans to thermocouple the packages. Installed in my Dynahi today (left channel shown) and did some bias checks. All look good. The difference between left and right is just a bit perceptible

I’ll fabricate some adapter board with protoboard if feasible so I can heatshrink the packages back to back. If can’t I’ll just use the included metal cans with some thermal paste. 

[penmarker]

Just found these k849/j689 adapter boards with incompatible traces from last time's group buy. I could probably adjust these instead to make them pin compatible with back to back J74/k170

[Pars]

Is the Dynahi setup with 14 pin DIP sockets for the input devices (ie, THAT340)? If so, you can just plug the JFETs into the DIP sockets (assuming you use good machine pin DIP sockets). No adapter required.

Edited March 10, 2023 by Pars

[penmarker]

3 hours ago, Pars said:

Is the Dynahi setup with 14 pin DIP sockets for the input devices (ie, THAT340)? If so, you can just plug the JFETs into the DIP sockets (assuming you use good machine pin DIP sockets). No adapter required.

Yup, I'm using Mill Max sockets for the THAT340 so swapping them out was quick. I was thinking about coupling them together so each pairs will be same temperature.

Left it on for a while and touched the jfets but it seems they stayed cool. Is their current draw way lower than what we would find in speaker amps?

[penmarker]

After replacing the input devices with K170/J74 I'm still hearing noise from the single ended out (with sensitive headphones). It sounds sort of like a frying sound that comes and goes. I probed the amp with my USB scope. The signal from FET devices and the first transistor is very clean (Q4, Q5).

I found the noise came from the transistor pairs before the power transistors.

Not sure what causes this. Is it because they're not closely matched? Is it dirty power?

Edited March 11, 2023 by penmarker

[MLA]

Since you're using a GRLV, dirty power would be unlikely I think.

I would replace those transistors for starters. Mpsa06/56 are in stock at least at Mouser.

[penmarker]

On 3/12/2023 at 3:27 AM, MLA said:

Since you're using a GRLV, dirty power would be unlikely I think.

I would replace those transistors for starters. Mpsa06/56 are in stock at least at Mouser.

Thanks. I’m currently using MPSW parts from eBay. My assumption is they’re genuine but it’s been years ago so I can’t check anymore. Bought as a lot for 06 and another lot for 56. I matched them by HFE with a diy transistor tester.

[MLA]

They are probably genuine if it was long ago, but may not necessarily have been new/unused. Anyway, replacing those and seeing if you get a clean signal could be an easy first check.

Where do you connect to ground for the SE output? Needs to be from signal ground if I remember correctly; Kevin had some very clear instructions on that early in the thread.

 

Edited March 13, 2023 by MLA

[penmarker]

The ground for SE output goes to one of the output ground. All the grounds on the board are on the same plane. For SE in you'll need to tie the L- and R- to their respective grounds.

[MLA]

On 7/6/2015 at 12:08 PM, kevin gilmore said:

for signal input, unbalanced you put hot to amplifier input +

you put ground to signal ground and input -

 

for output you wire hot to amplifier out +

and you wire ground to signal ground

 

anything else, and bad things happen

 

[Pars]

I know Amb recommended that for the SE out, the - goes back to the PSU ground, not the amp board grounds. That is the way I always do it, and never had problems. They are of course at the same potential as the amp board ground is, but gets any noise further away from the amp boards, at least that was his rationale for doing it that way.