I think what ANADIGIT meant in his post was that the compound act as a heatsink? Or is it about the real heatsink?! Anyway, I post below his statement because it is a good reference...
Hi NagysAudio
If you already have a Goldmund amp and the schematics are the same. I don't understand why you want to build this amp.
If this Goldmund amp is better than your Goldmund amp, then this suggests the magic is not in the amp circuit, but somewhere else, eg grounding, shielding, power supply etc.
So if the improvements are not in the circuit, can we really reproduce what Goldmund has done?
If you already have a Goldmund amp and the schematics are the same. I don't understand why you want to build this amp.
If this Goldmund amp is better than your Goldmund amp, then this suggests the magic is not in the amp circuit, but somewhere else, eg grounding, shielding, power supply etc.
So if the improvements are not in the circuit, can we really reproduce what Goldmund has done?
Here's Goldmund Mimesis 6:
Nagys Audio - Products - Provider of High End Audio Interconnect Cables
This is my personal amplifier. Nagys Audio is my company.
Nagys Audio - Products - Provider of High End Audio Interconnect Cables
This is my personal amplifier. Nagys Audio is my company.
apexaudio - Of course it's overpriced. It's severely overpriced! Just like all other audiophile equipment. But a lot of the overpriced junk, even from reputable companies sound like garbage, at least Goldmund is the exception. Goldmund's sound is extremely addicting. It's hard to listen to anything else afterwards. And that's true for all of their components, especially speakers. The amp has an extremely fast bass response and an extremely lively/bright/open treble/midrange. And it does that without any hint of grain, or harshness, or shoutiness, or glare. The sound just lingers right in front of you. It's hard to explain it and put it into words. It's something one has to experience. And I'm not in any way affiliated with Goldmund, so this is just my enthusiasm.
There's no need to clone the PCB 100%. I don't want to use the modules. It's extra work and expenses. If anything, Alex's PCB might even be better designed. The input stage is right there on the PCB itself. I like it better that way and it's easier to build.
This amp shouldn't cost more than $700 if both channels are built into the same chassis. At that price, this is a no brainer.
There's no need to clone the PCB 100%. I don't want to use the modules. It's extra work and expenses. If anything, Alex's PCB might even be better designed. The input stage is right there on the PCB itself. I like it better that way and it's easier to build.
This amp shouldn't cost more than $700 if both channels are built into the same chassis. At that price, this is a no brainer.
D2, D5, D4, D3, and T6 are all incorrectly connected. I know this amp and this circuit and I can promise all potential builders that it will not perform correctly. Beware.
The diodes can be reversed/changed while building the amp, no big deal. No changes are necessary to the PCB.
However, T6 cannot be easily rotated on the existing PCB. The PCB's tracks would have to be modified around T6.
EDIT: Actually, T6 can be rotated 180 degrees and its legs can be manipulated to make it still work. Kind of silly, but it would still work correctly. I'll still do the group buy. Anyone who connects the diodes and T6 the way it's shown in this schematic, please do not come to me for help. To make your amplifier operational.
The diodes can be reversed/changed while building the amp, no big deal. No changes are necessary to the PCB.
However, T6 cannot be easily rotated on the existing PCB. The PCB's tracks would have to be modified around T6.
EDIT: Actually, T6 can be rotated 180 degrees and its legs can be manipulated to make it still work. Kind of silly, but it would still work correctly. I'll still do the group buy. Anyone who connects the diodes and T6 the way it's shown in this schematic, please do not come to me for help. To make your amplifier operational.
Last edited:
How about that - they really do have that transistor connected upside-down! (as seen in the photo)
An interesting thing about BJTs is that if you swap the collector and emitter of an NPN transistor it still works as an NPN transistor, but with very different characteristics. When they're diode-connected, it shouldn't make much difference.
To test that theory, I simmed the simple circuit attached below, with the results shown.
The difference is small but presumably there's some good reason Goldmund chose to do it the way they did.
An interesting thing about BJTs is that if you swap the collector and emitter of an NPN transistor it still works as an NPN transistor, but with very different characteristics. When they're diode-connected, it shouldn't make much difference.
To test that theory, I simmed the simple circuit attached below, with the results shown.
The difference is small but presumably there's some good reason Goldmund chose to do it the way they did.
Attachments
godfrey (or anyone else kind enough) - Can you please do a quick simulation using -60V and using the T5, T6, D1, R11, and R12 from the latest schematic that Alex posted? Post #549. If you can, please do two separate simulations, with T6 connected both ways (Base and Collector shorted and 6V Zener diode connected to the Emitter. And Base and Emitter shorted and 6V Zener diode connected to the Collector). You can substitute any NPN transistor for T6 and T5, please make them the same (You can use BC182B, BC449, BC546B, MPSA93, MPSA92, etc.). Please show voltages at T6's Emitter, T6's Collector, T5's Collector, and T5's Emitter in both simulations. If the voltage difference is substantial, then it might unbalance the circuit. Can you also, if possible, please post the current and any gain changes. This should put this argument to rest. Thank you!
Last edited:
I did a simulation using the following parts: T5, T6, D1, R11, and R12 from the latest schematic that Alex posted, post #549. I grounded resistor R11.
If connected the correct Goldmund way (T6's Base and Emitter shorted and 6V Zener diode connected to the Collector). I get the following data on T6:
Reverse Active
Ic = -1.6 mA
Ib = 800.65 uA
Vbe = 0 V
Vbc = 587.42 mV
Vce = -587.42 mV
If connected the wrong way, as apexaudio suggested (T6's Base and Collector shorted and 6V Zener diode connected to the Emitter). I get the following data on T6:
Forward Active
Ic = 1.59 mA
Ib = 15.85 uA
Vbe = 587.42 mV
Vbc = 0 V
Vce = 587.42 mV
NOTE: Every other parameter stays the same in the schematic regardless of how T6 is connected.
If connected the correct Goldmund way (T6's Base and Emitter shorted and 6V Zener diode connected to the Collector). I get the following data on T6:
Reverse Active
Ic = -1.6 mA
Ib = 800.65 uA
Vbe = 0 V
Vbc = 587.42 mV
Vce = -587.42 mV
If connected the wrong way, as apexaudio suggested (T6's Base and Collector shorted and 6V Zener diode connected to the Emitter). I get the following data on T6:
Forward Active
Ic = 1.59 mA
Ib = 15.85 uA
Vbe = 587.42 mV
Vbc = 0 V
Vce = 587.42 mV
NOTE: Every other parameter stays the same in the schematic regardless of how T6 is connected.
I woke up and had 2 pages of reading to do! I'm glad Godfrey indicated the symmetrical nature of a transistor, yes it will work the same with a slightly different operating point. The question to ask however is whether SPICE pays enough attention to reverse transistor behavior. There are parameters for it, but the vast majority of users are more interested in the conventional use. Also, Ccb may be different than Cbe.
I would have simulated it the same way, except I thought the original schematic had a mistake. Will change again.
In any case, I don't think the CCS is that important. I have already indicated that with how little variation this part of the circuit sees, it's contribution to overall specs will be small compared to what the FETs do. Second, it looks to my like this CCS was designed for low tempco, and this is smart considering the FETs use just a resistor for setting the bias right. I think we can "get away" with doing it like Goldmund.
Regarding forks of the original design... While we might make improvements, it would no longer be the "goldmund" amp, as I believe Nagys has made clear. Rather it would be more like creating a modern design using the Goldmund as a starting point; we already know it sounds good.
- keantoken
I would have simulated it the same way, except I thought the original schematic had a mistake. Will change again.
In any case, I don't think the CCS is that important. I have already indicated that with how little variation this part of the circuit sees, it's contribution to overall specs will be small compared to what the FETs do. Second, it looks to my like this CCS was designed for low tempco, and this is smart considering the FETs use just a resistor for setting the bias right. I think we can "get away" with doing it like Goldmund.
Regarding forks of the original design... While we might make improvements, it would no longer be the "goldmund" amp, as I believe Nagys has made clear. Rather it would be more like creating a modern design using the Goldmund as a starting point; we already know it sounds good.
- keantoken
Here you go - anything to fuel an argument! 
I tried a few different transistors and got suspiciously similar results. In each case the difference in R3's voltage is only about 0.1%, which will be insignificant compared to the differences caused by resistor and zener tolerances, component-to-component variation etc.
On the other hand maybe my sim software just got it wrong. That wouldn't surprise me either.
If there is a significant advantage one way or the other, it may be due to some other consideration e.g. noise or tempo (which I can't sim).
I tried a few different transistors and got suspiciously similar results. In each case the difference in R3's voltage is only about 0.1%, which will be insignificant compared to the differences caused by resistor and zener tolerances, component-to-component variation etc.
On the other hand maybe my sim software just got it wrong. That wouldn't surprise me either.
If there is a significant advantage one way or the other, it may be due to some other consideration e.g. noise or tempo (which I can't sim).
Attachments
Nagys, for the original version you should record Ie rather than Ic, since the LTP sees Ie. I found in simulation that with the 2N5550, Ib was .5mA, making the output 1.5mA rather than 2mA like it was before. Reverse Beta will vary for different transistors, so it may be important to use the BC182B here. For instance, changing to the 2N5210 I get .8mA Ib, for the MPSA18 I get .3mA Ib. This change also has a large effect on the output stage bias. Using the 5210 increases crossover distortion by a large margin by reducing output stage bias.
- keantoken
- keantoken
Okay, I just did a big simulation on temperature.
Both versions of the original CCS have a POSITIVE temperature coefficient. The standard two-transistor CCS has a NEGATIVE temperature coefficient.
The "emitter-down" version has the smallest tempco. The "emitter up" version has worse tempco. The two-transistor CCS has the worst tempco, and it is also a negative tempco. These results were found the same for 3 different transistor models.
It seems to me Goldmund's biggest concerns were temperature and noise. Sometimes a reversed transistor is actually quieter.
It is also possible that Goldmund made a "benign mistake" perhaps, which was never corrected because it wasn't a big problem? It works the same. I will simulate tempco of the Goldmund amp using all 3 CCS types.
- keantoken
Both versions of the original CCS have a POSITIVE temperature coefficient. The standard two-transistor CCS has a NEGATIVE temperature coefficient.
The "emitter-down" version has the smallest tempco. The "emitter up" version has worse tempco. The two-transistor CCS has the worst tempco, and it is also a negative tempco. These results were found the same for 3 different transistor models.
It seems to me Goldmund's biggest concerns were temperature and noise. Sometimes a reversed transistor is actually quieter.
It is also possible that Goldmund made a "benign mistake" perhaps, which was never corrected because it wasn't a big problem? It works the same. I will simulate tempco of the Goldmund amp using all 3 CCS types.
- keantoken