Jim's Audio Goldmund/Symasym clone

[linuxguru]

This thread is for the discussion of EBay seller Jim's Audio Goldmund/Symasym kit - which is a single-board amp with rectifier, filter caps, 2 channel power amp and uPC1237 speaker protection circuit all on a single compact PCB. All matters pertaining to the construction, modification, simulation, component substitutions, and anything else pertaining to this kit may be discussed in this thread.

Here's a simplified simulation schematic of one channel of the Goldmund/Symasym, with a few modifications.

1) The GNFB resistor network has been changed to 680/27k instead of 1k/22k supplied in the kit - this improves the phase margin and stability.

2) An additional inner feedback loop has been added from the output of the driver stage (330k), which helps to improve phase margin as well as imaging.

The second pic is the simulated small-signal loop gain, showing a phase margin of ~80 degrees at unity loop gain.

The third pic shows the THD20 FFT at 3.5W output into 8 ohms, showing a pleasant harmonic cascade, indicating warm-sounding audible sonics.

[linuxguru]

Earlier discussions on the Goldmund/Symasym kit can be found on the thread relating to the Yuanjing Gainclone in the chipamps forum, starting at this post:

"Yuanjing" Gainclone 3886 - eBay amazing value ?

Note that the Yuanjing thread contains posts pertaining to both the Yuanjing LM3886 gainclone as well as the Jim's Audio Goldmund/Symasym kit. The aim of the present thread is to collect all future posts related to the Goldmund/Symasym kit here.

[barry.childs]

I put some pics on the other thread. So here they are again

[barry.childs]

That's a very nice diagram. I could hardly see the original. Can you please explain the 330k feedback loop and can I do the mod without removing the board? Do you know what the input impedance is and would there be a big benefit by adding a pre-amp stage?

[AndrewT]

Originally Posted by linuxguru

Earlier discussions on the Goldmund/Symasym kit can be found on the thread relating to the Yuanjing Gainclone in the chipamps forum,

that explains my confusion when I saw the discrete output devices in a 3886 thread.
Thanks for the split.

BTW,
I think C3 is far too big.
I would try values around 1uF.

[barry.childs]

Sorry Andrew, that was my fault due to me and linuxguru both building both amps at the same time.

[linuxguru]

Originally Posted by barry.childs

Can you please explain the 330k feedback loop and can I do the mod without removing the board? Do you know what the input impedance is and would there be a big benefit by adding a pre-amp stage?

The 330k feedback loop helps linearize the Class-A stages of the amp, i.e. the LTP, the Hitachi-type VAS and the Self Type-II driver stage. It also contributes a few degrees of phase margin and as a bonus, helps to improve imaging. The downside is that odd harmonics tend to increase slightly (actually, that problem is an artifact of the 33 + 33 ohm resistors between the emitters of the drivers - 47 + 47 ohms will greatly reduce the odd harmonics and also make the drivers run cooler).

This 330k loop is optional as it stands now, but helps a lot in higher loop-gain variants which I have in simulation, where the degeneration resistors in the LTP and VAS are smaller. I'll post a schematic of those variants after optimizing the values a bit for stability.

The input impedance is fairly easily obtained by inspection: assuming a very high gate impedance for the 2sk170 JFET, the input impedance is mainly 100k, i.e. the input shunt resistance.

Edit: I don't think it needs a pre-amp, unless you want to add a Baxandall tone-control stage also.

Andrew: Yup, 10 uF is overkill for the input DC blocking cap, but that's what was supplied in the kit - Nichicon Muse BP 10 uF/50V. I wish there was more space there, to accomodate a tubular axial-lead film cap, but there's isn't sufficient space, even if it is mounted vertically. This location is also good candidate for a small Black Gate, Elna Cerafine or Nichicon Muse KZ, say 4.7uF/50V or similar.

[barry.childs]

Thanks Linuxguru, It would appear then that at aprox 100k input impedance any normal low level device such as cd, tuner etc. would be bridged and therefore not require a pre-amp at all. I do not want tone controls. I never did like them except for old records and I don't use them anymore. I only have selector and volume (I might fit a ladder type attenuator, but i'm not sure there is any advantage).
edit: did you consider fitting a more suitable DC blocking cap underneath the board?

[linuxguru]

Originally Posted by barry.childs

did you consider fitting a more suitable DC blocking cap underneath the board?

Not yet - that MUSE BP seems to be good enough for now. However, a Cerafine, Black Gate or MUSE KZ remains a viable experimental alternative.

I swapped out the 33 + 33 ohms resistors between the emitters of the drivers and put in 47 + 47 ohms, 0.5W instead. I also added the 390k inner-feedback resistors to the underside of the board with insulating sleeves (it's possible, but messier, to do this on the component side also, without removing the board).

In the sims, anything from 100k to 1M and higher works here - lower values will increase H3 and make it edgy and sharp. Higher values are more laid-back and warm. I just started auditioning it with 390k resistors, and my initial impressions are that it seems more focussed, sharply imaged and darker than without the inner feedback loop. Piano pieces are audibly more forward and finely imaged with the inner loop. Some experimentation may be required to get the optimum value .

I would regard the change of the driver emitter resistors to 47+47 as more desirable than the use of the inner feedback loop, from the simulation numbers alone. However, YMMV - I did not audition the two changes separately.

[snip6600]

Hi all, I guess this is my first post on diyaudio.

Firstly thank you to LinuxGuru for investing so much time in perfecting this kit.

I have gone ahead and ordered one of these suckers.

I plan on sourcing original Toshiba output devices and will implement at the time of construction the so far suggested mods by Linuxguru;

Changing the resistors in the feedback loop to 27k/680R
The VAS rail caps from 100uF to 330uF
The DC blocking cap in the NFB network to 180uF Polymer

Also I plan on changing all the small ceramic capacitors for silver mica and will likely change signal passing resistors for Kiawe's.

Linuxguru: is there any chance you could highlight the components you have changed/added by changing them to a different colour for instance? I think that would be quite neat.

Also, those of you who have built this already, what size heatsinks have you gone with?

Best regards
Jan