The Very Best Amplifier I Have Ever Heard!!!!

That is why I took the time to build a variac front end for the output power supply for the work I am doing on the KO amp.
:headbash: No one ever listens to me !! Power up (run the amp) with just the drivers first. For my lateral experiments , I used 2- 100R RE's , took my NFB off the middle of them. Between the driver emitters , I had 2V. hooked up the laterals ...10-20ma bias ... adjusted, DONE.After this first step, on multi pair OPS's ,I "fire up" just 1 pair of either laterals or BJT's first , adjust to proper current ...populate the rest.

As a added safety factor , I use 2- 1/2w/ 10R resistors to power the voltage stage/driver initial test. NEVER have I blown an amp entirely. 🙂

OS
 
OK Huston I got a problem? I fire it up (took Zen advice and go for a 2SK1530) measure voltage drop across source resistor and value 0.0. Changed R22 from 330 to 270. I don’t have tripot, value the same 0,0 Liliya wrote something that he changed R19 that’s R20 for 9.3 and adjusted bias. So any Ideas?
 
This is what I have "learned" today about my Goldmund clone.

Some of this adwises are perhaps "baby food"and "to simpel" for some of you, but for some inexperienced it may be okay with some practical advice.

There is is no problem to set the bias by using a potentiometer. I desoldered and liftet one leg of R20, let the resistance of 330 ohms to sit as a form of security. Between the free leg of this resistance and the "empty" eye on the PCB, I mounted a multiturn pot value 1 Kohm. A multiturn makes it possible to increase the bias gradually, and when the adjustment is completed, give a fairly accurate indication of how big the final fixed resistor should be.
I have so far landed at 135mA on each source resistance.

Setting the bias influence on the serieresistors R31/32.

R31/32 is placed on the PCB where there is little space, and where it is difficult to solder without having to turn the PCB (and dismantle PCB from the heat sink). It may be wise to think about this during soldering and before you do all settings/testing.

This will "easy" the installation of R31/32 and make it possible to solder from the component side of printed:

Create a "serie value "of R31/32 with two resistors. Solder these resistors standing and connect them with each other. When you have to change the total resistans value, change one of them, which you now do easily from the compont side.

With me, drivers T11/12 get hot. To hot, I think. First I choosed to use
MPSA 42/92 because I had these on hand. The drivers need, in my experience, cooling, or they will after some time "retiring" because of the heat. MPSA 42/92 does not have cooling possibilities.

The recommended drivers SS71/74 can be mounted with small heat sink.
I recommend to use those for drivers. They fit of course on the PCB.

Although, I chose to use some 150 V types. They were on hand. These are types of TO220 (BCE), and of course, fits very poorly into PCB, but with a little patience, this can be fixed. I expect that the drivers I use are of very little importance for listening result.

Work will continue to morrow.


Eivind Stillingen

Liliya what value did you get for resistor + trimpot in series (R20)I presume for yours biasing?
 
Josip-Del

In an earlier post I have written:

There is is no problem to set the bias by using a potentiometer. I desoldered and liftet one leg of R20, let the resistance of 330 ohms to sit as a form of security. Between the free leg of this resistance and the "empty" eye on the PCB, I mounted a multiturn pot value 1 Kohm. A multiturn makes it possible to increase the bias gradually, and when the adjustment is completed, give a fairly accurate indication of how big the final fixed resistor should be.
I have so far landed at 135mA on each source resistance.

I measured the total resistance for 135mA to be 391 Ohm(resistor + pot) so I choosed the nearest value: 390 Ohm.

Eivind Stillingen
 
Last edited:
Here is a photo of the power unit I use for the front end stage to my Goldmund clone.

The idea how to connect another transformer to the one I already have, is taken from an earlier built I did some years ago: LFA 50 class A amplifier.
I use a 2x20 V AC transformer to each channel to get 2x80 V DC out to the front ends(and to stand lower than 100V DC).

You can find more info here how to connect the transformer:
http://www.jls-info.com/julien/audio/Schema_Alim_LFA.gif

I have dismounted the original power to the front end. Many of this components can be re-used into MJL 21193s power unit.

Eivind Stillingen
 

Attachments

  • 2011_02190002.JPG
    2011_02190002.JPG
    80.5 KB · Views: 1,625
  • 2011_02190003.JPG
    2011_02190003.JPG
    99.5 KB · Views: 1,593
jacco_vermeulen, you wrote in # 1174:

In case you run into stability issues, ground the heatsink and solder the C's at the drains to the heatsink. (in the Hitachi appl. note, btw).

From the heatsinks: Do you mean 1) to ground on the PCB or 2) to the point where I ground the "hole" amplifier to the chassis (in my case, by the phone socket for input)?

Solder C to the heatsinks:
If you look at Alex PCB and how C are connected to the drains, how can this be done in a practical way? They are now connected between Drain and Source.

I consider to use both this recommendations from you, just in case.

Eivind Stillingen
 
Last edited:
With good help from MJL 21193 to do some component changes to his power unit so it fit my needs, I have now finished the work. With this power unit I am able to ajust the front end voltage to exactly +/- 80 V DC.
This also stabilize DC offset on speaker out.

MJL have made some measurement. This unit, I think, outperforms by a wide margin the original. You can read about the unit here: http://www.diyaudio.com/forums/solid-state/180749-patchwork-6-power-supply-upgrade.html

After this changes, I am looking forward to listen to the Goldmund later to night

Eivind Stillingen
 
With good help from MJL 21193 to do some component changes to his power unit so it fit my needs, I have now finished the work. With this power unit I am able to ajust the front end voltage to exactly +/- 80 V DC.
This also stabilize DC offset on speaker out.

MJL have made some measurement. This unit, I think, outperforms by a wide margin the original. You can read about the unit here: http://www.diyaudio.com/forums/solid-state/180749-patchwork-6-power-supply-upgrade.html

After this changes, I am looking forward to listen to the Goldmund later to night

Eivind Stillingen

But of course , I commented on the infantile voltage doubler/cap mult. from the very beginning. Anything would outperform the "profit margin influenced" original design. A GOLDMUND deserves far better. 🙄 😀

OS
 
Ostripper

Yes, you are (were) absolutly right. I took your "warning" from the first time you wrote about this weakness, and I have been looking around to find a suitable power to handle such high voltage.

Last night I was listening to my Golmund and I must say again: The way this amplifier play the upper octave, has never been done in my house before (and I have own and listen to a lots of gear during the last fourty years). This was also my main reason to jump into this risky procject. "The very best" does not excist (in advertising,yes) but different amplifier have different strenght and weakness. I have read about this amplifier before building it (The Absolute Sound among others) and the oppertunity to own a Goldmund within a reasonible price (yes, I agree, all Goldmund equipment are heavy overpriced) challenged me. The Goldmund clone will soon "take over" the place to handle frequencies from 600 Hz and up to the top in my Magnepan 1.6. I run this loadspeakers from a heavily modded DCX 2496 trough NAD 208 (subwoofer), a Nelson Pass F4 (50-600 Hz) and a modded MA tube amplifier for the top.

Talking about stability and how this amplifier measure, there is still a job to do. I will wait and hear what other builders have experienced.

jacco vermeulen gave some advise about stability in an earlier post ( 1174). Do you have any comments on that? Read my #1453.
I consider to do this mod, but are unsure of what he means about the location of capacitors, which now sits between drain and source.

Eivind Stillingen
 
jacco vermeulen

Thank you. I have read the thread that you referred to.

Hansen wrote: Make sure you are using gate "stopper" resistors as close to the gates as physically possible. As a starting point, these should be somewhere in the range of a few hundred ohms to a few thousand ohms.

Assuming that you are using source followers in the output stage, another point that can help with oscillations is to run a small cap (1000 pF or so) from the drain of each device to the heatsink itself. This idea was taken from the Hitachi application sheet, and will stop oscillations in many cases. It is essentially a local bypass cap on the power supply rails to the closest chassis ground, so it won't hurt the sonics at all.


Gate stoppers: I use 100 Ohm. I know that increasing the value is one way to cure oscillation. Drawback: FETs get slower.

1000 pF from each drain to the heatsinks can easily be done in my Goldmund, but what about the capasitors that already are mounted between drain and source. As far as I have understood this are placed there, not to cure oscillation, but to equalize capasistance to the FETs. Shall they stay, or shall I remove them.
May bee this is a silly question, but Hanson wrote "assuming you are using source followers". When it comes to Golmund, is it???? I presume you know the original schema. I am not familiar with "source follower".

Changing the bias is also mention as a cure in this thread. I went for a bias current of aprox. 140 mA to each device, but have not looked at a scope to see if this (bias) influence on possible oscillation. My FETs are Exicon 10P16/10N16.

Can oscillation in any way be easily registrated whitout a scope?

I have registrated the following: Without any connection to a preamplifier or to speakers, heatsinks get warmer than what happens after hours of listening on relativly high level ( I stayed up to night, listening, difficult to go to bed).
Oscillation????

As you can read from my post to Ostripper, Goldmund clone play(for my ears) music in a very satisfying way.

Eivind Stillingen
 
Last edited:
Eivind,

only (bypass) caps i've seen in the schematic/boards are 10nF between drain and (star) ground, not the source of the power MOSFETs.

Common ways to compensate for difference in input capacitance are :
- adding caps in the hundred pF range between gate and source
- using different value gate resistors (e.g. 390R and 680R in the Tandberg power amp i linked to)

Only relevant part of Mr Hansen's post to explain what i said in no 1174 :
another point that can help with oscillations is to run a small cap from the DRAIN of each device to the HEATSINK ITSELF. It is essentially a local bypass cap on the power supply rails to the CLOSEST CHASSIS GROUND.
 
Last edited:
The heatsink grounding and decoupling.
The extra cap being referred to must go from the supply to the sink.
An SF (=BJT EF) output device has the Drain attached to supply so an SF requires the drain to be capacitor coupled to the sink.

A common source output device (CFP) has the source connected to the supply, usually via a source resistor.
If the source resistor is in place, which side of it gets capacitively coupled to the sink. The supply side of the source resistor or the source of the output device to the sink?