Wednesday, September 3, 2014

Report on a "Goldmund Clone" amplifier from eBay

I recently bought a new power amplifier from a seller on eBay. It was sold as "HiEND power amplifier 195W*2 8ohm ,aluminum,base on Gold*Mund mimesis 29" The amp is rated at 195W into 8ohms, and uses 8 lateral MOSFETs per channel.

The Goldmund design is quite well-known among DIY amp builders. It was said to produce a very good sound quality. However, even the original Goldmund amps (some of them, at least) had a tendency to self-destruct. And the seller seemed to be warning about possible problems - here was his list of warnings:
1) do not only connect one speaker..
2) do not use tube pre-amp.
3) if you want change the signal or the speaker cable,you must turn off the power amp  first
4) when you  playing very very loud sound ,please  touch the heatsink and be sure the heat sink  no too hot so that can cook the egg..

So I went into this with my eyes open. One reason was price - about us$550 including shipping. I asked why no tube preamp, and got no answer. 

I was impressed when it arrived.

All-aluminium case. Large transformer. Speaker protection board. 16 lateral mosfets - these alone cost more than $10 each. It uses 2SK1058 and 2SJ162 MOSFETS. OK, not the original Hitatchi, but lateral mosfets are much better for amplifiers than the cheap switching MOSFETs often used.

And when I plugged it in to play, wow, it sounded great.


For a few hours. It actually died while it was idling. Had been on for about 8 hours. One MOSFET blew and took the fuse with it. So I happened to have a few spares handy. 

I figured that the gate-stopper resistors were too small, so I replaced these with larger values - 470R and 820R (different values for the P and N channel MOSFETs, as they have different input capacitance). Gate stoppers are resistors that are meant to stop oscillation, which is a common problem with MOSFET output stages.

So that didn't work.

I spoke to the seller and he admitted that quite a few of the amps he sells have problems, but he claimed that this one had been running fine, in Hong Kong, for a week or more. I noted that HK has 220VAC. My voltage is 240V, so significantly higher. I wondered if this might be the problem.

I decided to try reducing the voltage that the amplifier saw. I used a trick that John Broskie taught me, using a 25 Volt transformer to subtract 25 volts from the 240VAC that the primary sees; this reduced the input voltage to 215VAC, which meant that the rail voltage dropped from 68-0-68 to 58-0-58

This solved the problem. The amp ran just fine now, and still sounded great. I also obtained a circuit diagram of the variant used in this amp. Thinking about what had happened, I figured that the problem was due to the voltage stage getting too warm, which caused it to oscillate at some point, which then caused a MOSFET to die, as the rail voltage was quite close to the MOSFET's max voltage of 140V. 

I tested this theory by making a simple regulator (using 56V zener as shunt) to power the voltage stage. The output and driver stages were still on 68-0-68. This also did not self-destruct, so it seems my theory was correct.

I plan to make a more sophisticated regulator soon, probably some kind of capacitor multiplier.

More Problems...

However, there was another problem with the amp. It uses 4 pairs of parallel MOSFETs per channel. And these are directly connected to the output. Most circuits use a low valued source resistor. This is designed to prevent one pair of mosfets taking all of the load current, using local feedback (degeneration). When Goldmund builds their amps, they match the MOSFETs so that this resistor is not needed. However this amp did not have matched MOSFETs I was sure. I did a high power test and sure enough, I managed to kill the amp again.

So I put in 0.1ohm source resistors. This made it much harder to blow the amp, but I still managed in the end.  Replacing them with 0.2ohm resistors solved the problem.

There you have it - a couple of small fixes and you have a great amp. I was amazed at how good this sounded. Compared to my main amp, the Moskido, a hybrid, non-feedback amp, I find the Clone amp to have more detail, speed, dynamics. But the Moskido sounds more musical. That's the magic of tubes..

And it works now just fine with a tube preamp! 

You Want One?

I'd be happy to explain how to make the changes needed to make this a functioning amplifier. I can also let you have the circuit diagram. I could even do the work for you myself but this would cost in shipping. But still a bargain considering that you'd need to spend a lot more than this to get an equivalent-sounding amp.

Black Discus on the Amplifier

As you can see, I've added a fair number of CanOpeners to the amp. I was trying them on the capacitors. The best place for them is over the heatsinks that cover the voltage amplifier transistors.