You solve the hum only by removing the trimmer and using a servo. A minimal imbalance will cause hum. But while experimenting the hum is good, and helps to understand the amp is alive. You better not modify the CCSs to have the same current and lower the 5.1V, their already low impedance would drop too much. You can try to use a TO220 DN2535N5 or DN2540N5 which has a Vgs.off around 1.5V (mine are all close to this value, buy some and make a selection). 1KOhm on the gate, 12Ohm on the source, and connect both resistors at their other ends. You obtain a 125mA CCS. This solution is asymmetrical (no P channel version for this mosfet) but works. The 5.1V is now 1.5V.
Again take a look at my post n.2107, it has the same topology you need, but the CCSs have a far lower value of 12.5mA because it is enough in my schematic.
Again take a look at my post n.2107, it has the same topology you need, but the CCSs have a far lower value of 12.5mA because it is enough in my schematic.
From what I can see in sims, removing the offset trimmer does not seem to improve PSRR, for some reason it actually seems to get slightly worse removing it..?
It also seems my sim was pretty realistic. I compared the sim with basic voltmeter AC measurements today, and they were pretty close. I have abt 3mV hum on the output of the amp, and I have confirmed it's from PS ripple.
I simulated a cap multiplier for the front end excluding drivers/CCS and output transistors, and that looked like a significant improvement. However, when I started looking at the board, it's a bit hard to split the supply traces there and insert a cap multiplier, so I tried to sim it with the drivers/CCS included in the cap multiplier supply. Performance improved a bit, but then there is the problem of dimensioning the cap multiplier for the higher current/power and TO-92 transistors are no longer sufficient. There is also the matter of reducing output swing even more with the voltage drop of the cap-mx plus the CCS. I need to think about it some more..
I think I will leave the CCS alone, 25V peak should be enough for my needs, no need to get greedy
It also seems my sim was pretty realistic. I compared the sim with basic voltmeter AC measurements today, and they were pretty close. I have abt 3mV hum on the output of the amp, and I have confirmed it's from PS ripple.
I simulated a cap multiplier for the front end excluding drivers/CCS and output transistors, and that looked like a significant improvement. However, when I started looking at the board, it's a bit hard to split the supply traces there and insert a cap multiplier, so I tried to sim it with the drivers/CCS included in the cap multiplier supply. Performance improved a bit, but then there is the problem of dimensioning the cap multiplier for the higher current/power and TO-92 transistors are no longer sufficient. There is also the matter of reducing output swing even more with the voltage drop of the cap-mx plus the CCS. I need to think about it some more..
I think I will leave the CCS alone, 25V peak should be enough for my needs, no need to get greedy
Today was a FAIL, cap multiplier on the front end did basically nothing to the hum..
I would like to avoid a servo, I have simulated a cap between the feedback resistors and ground that reduces the gain to 1 at DC, but some bias to the input stage is still needed I think.
Will see what I try next.
I would like to avoid a servo, I have simulated a cap between the feedback resistors and ground that reduces the gain to 1 at DC, but some bias to the input stage is still needed I think.
Will see what I try next.
I tried the 'brute method' today by doubling the PS capacitance. Hum was reduced to acceptable level, but can still be heard with ear close to the woofer.
I did some more listening, and I concluded I don't regret building it. What stands out to me with this amp subjectively is the treble, airy, spacious and clear. Mid is nice too but low end is not as good as a 'standard' amp with a lot of global feedback. Possibly the bass improved slightly from the added PS capacitance, but it's still lacking, as I expected. In general what also surprised me is the 'attack' in transients, like snare drums. I'm still running without the base diodes, as I don't see any reason to mount them.
Overall an interesting experience, -building a terribly measuring amp, and liking the sound from it.
I might try cap multipliers on the PS too (since I have a pair from another build), but I'm a bit concerned about their output impedance which is a lot higher than even poor capacitors, and it seems to vary with load too. I would like to avoid throwing loads of expensive PS caps into this build, especially since my intention is not to use it for bass.
I did some more listening, and I concluded I don't regret building it. What stands out to me with this amp subjectively is the treble, airy, spacious and clear. Mid is nice too but low end is not as good as a 'standard' amp with a lot of global feedback. Possibly the bass improved slightly from the added PS capacitance, but it's still lacking, as I expected. In general what also surprised me is the 'attack' in transients, like snare drums. I'm still running without the base diodes, as I don't see any reason to mount them.
Overall an interesting experience, -building a terribly measuring amp, and liking the sound from it.
I might try cap multipliers on the PS too (since I have a pair from another build), but I'm a bit concerned about their output impedance which is a lot higher than even poor capacitors, and it seems to vary with load too. I would like to avoid throwing loads of expensive PS caps into this build, especially since my intention is not to use it for bass.
The hum is injected via the trimmer. Try to decouple the noise by putting two 1uF very low ESR capacitors from mass to both ends of the trimmer, common points with 510KOhm resistors.
Just a try, because the capacitors will affect the input impedance, and you cannot leave them in place that way. If successful, to avoid messing up with input impedance you must insert a 100KOhm or greater resistor from the center of the trimmer to the first transistors' bases.
The original project must have a not-reported solution to avoid this problem or a very picky selection of transistors was made. Following the "original" schematic you find in this thread there is no solution to the hum using normal transistors you can buy from the best dealer.
When I removed the trimmer, the hum disappeared. But I was to use a servo, and no Cinput, I always have output capacitors on all preamplifiers I build.
Looking at your board, I see Q5 and Q6 (nomenclature from the original schematic) put on the common aluminum board. They do not warm up at all, so you better lift them to avoid thermal contact. The best would be to substitute them with more suitable devices for the purpose, like I did.
Just a try, because the capacitors will affect the input impedance, and you cannot leave them in place that way. If successful, to avoid messing up with input impedance you must insert a 100KOhm or greater resistor from the center of the trimmer to the first transistors' bases.
The original project must have a not-reported solution to avoid this problem or a very picky selection of transistors was made. Following the "original" schematic you find in this thread there is no solution to the hum using normal transistors you can buy from the best dealer.
When I removed the trimmer, the hum disappeared. But I was to use a servo, and no Cinput, I always have output capacitors on all preamplifiers I build.
Looking at your board, I see Q5 and Q6 (nomenclature from the original schematic) put on the common aluminum board. They do not warm up at all, so you better lift them to avoid thermal contact. The best would be to substitute them with more suitable devices for the purpose, like I did.
I think this is why a larger input cap was beneficial to PSRR in sims as well, because as long as the output impedance of the source is low, the input will be AC-grounded via the source. I did increase the input capacitance for this reason, but obviously it's not enough. And even if I put a cap multiplier on the front end (where the resistors are connected) it was still the same. As it is now, the cap multiplier is still on one board, but not the other. Hum is the same on both, and adding PS capacitance decreased it in the same way on both channels, so I'm thinking it must be coming from the drivers/outputs, but I'm not able to simulate it properly. I'm now at abt 4x35mF in the PS, and hum is audible close to the woofers.
Maybe I'll just slap on more PS caps and be done with it I still want to experiment a little with the base resistors too. I think this amp deserves a future built in an enclosure.
Do you think the little extra heat matters to those transistors? With the heat sinks I have, they get barely warm. I'm thinking I will leave it as it is, it works, and DC-offset is within reason. I think the front end is magnitudes 'cleaner' then the output stage, so that is where I'm looking to possibly do some small tweaks to see/hear what happens.
To me this amp is a bit like 'the bumblebee can't fly-theory'. It should not sound good according to objective measurements, but somehow (subjectively) it still does some things right. My curiosity around this was the reason for building it in the first place. I built many low distortion amps, and liked some, but definitely not all!
Maybe I'll just slap on more PS caps and be done with it
I still want to experiment a little with the base resistors too. I think this amp deserves a future built in an enclosure.Do you think the little extra heat matters to those transistors? With the heat sinks I have, they get barely warm. I'm thinking I will leave it as it is, it works, and DC-offset is within reason. I think the front end is magnitudes 'cleaner' then the output stage, so that is where I'm looking to possibly do some small tweaks to see/hear what happens.
To me this amp is a bit like 'the bumblebee can't fly-theory'. It should not sound good according to objective measurements, but somehow (subjectively) it still does some things right. My curiosity around this was the reason for building it in the first place. I built many low distortion amps, and liked some, but definitely not all!
Last edited:
I tried 24ohm resistors in parallel with the standard 27ohms today, but I actually found the sound a bit dull in the treble, but bass improved. I prefer the sound with the standard resistors (more 'airy' treble), since I will probably only use the amp for tweeters. I also put in the schottky diodes and did not notice any change in the sound, but that was expected since they would only conduct at really high output currents, and such high levels make my ears hurt anyway.
I solved the hum by using rectifier-16mF-capmx-18mFcaps (per rail, so *4). Now there is no hum, only a slight hiss from the tweeter when I put my ear next to it. I think next mission is to box it.
I solved the hum by using rectifier-16mF-capmx-18mFcaps (per rail, so *4). Now there is no hum, only a slight hiss from the tweeter when I put my ear next to it. I think next mission is to box it.
When I made 2 base resistors at 22 ohms in parallel, I got 11 ohms, the quiescent current of the output transistors increased. The music began to play and became much more interesting. The radiators now heat up 10 degrees higher, about 50 (maybe higher if you play louder for a long time), but I’m not afraid , in the power supply circuit there are ceramic temperature sensors at 65-68 degrees, they will turn off. Before the alteration it was a bit boring, now it competes with the A60 and there is no reason to say that this is bad. Graphs and measurements are good and you need to do it, but no one canceled the ears if there are 8 pairs of them.
There have never been any problems with highs, mids, lows, now it’s even better. Again we sat and listened and compared, all genres play perfectly.
There have never been any problems with highs, mids, lows, now it’s even better. Again we sat and listened and compared, all genres play perfectly.
I did another try with 10ohm base resistors today, but after some time playing the magic smoke escaped on one channel, and I had to replace the output transistors and the power transistors in the cap-mx. The base resistor on one of the output transistors was up in flames Luckily I was just next to the amp and disconnected it instantly. Now I'm back to the 27ohm resistors and it's working well.
I have only 30V rails, but also the output transistors are smaller package type than the originals (could not find them).
Luckily I was just next to the amp and disconnected it instantly. Now I'm back to the 27ohm resistors and it's working well.I have only 30V rails, but also the output transistors are smaller package type than the originals (could not find them).
The circuit is fully consistent with the 108 first version. All components are original, Takman Ray resistors. Power supply 47V DC. The base resistors are 11 ohms, the quiescent current has increased, the sound has improved. But for this you need to have large radiators for cooling, otherwise overheating and failure. I use the Chinese Dartzeel case. In the original amplifier, the power supply is 56 volts, so the base resistors are 27 om. By the way, it is described in the description It also gets quite hot.Read.2160-2180 What output transistors are you using?
Last edited:
You can see my heat sinks in the test setup here:
https://www.diyaudio.com/community/threads/dartzeel-amp-schematic-build-this.134362/post-7597746
They did not have time to get hot, it took only a few minutes for the amp to blow.
I simulated it before too, and it looked like it should work. I think I will leave it as is.
https://www.diyaudio.com/community/threads/dartzeel-amp-schematic-build-this.134362/post-7597746
They did not have time to get hot, it took only a few minutes for the amp to blow.
I simulated it before too, and it looked like it should work. I think I will leave it as is.
@Rallyfinnen
Did you buy your transistor from an official good dealer? The scene you described brings me back to when I bought transistors from eBay. They were all fake! The reason for the explosion cannot be a thermal avalanche, it takes time. It is an error of a fake component.
Did you buy your transistor from an official good dealer? The scene you described brings me back to when I bought transistors from eBay. They were all fake! The reason for the explosion cannot be a thermal avalanche, it takes time. It is an error of a fake component.
Yes, they are samples directly from Onsemi, so no doubts there! They are however NJW, not MJL as I mentioned before, so smaller package and lower heat dissipation, lower Cob, but very similar other than that. I also, mounted them with sil-pads in the proto-setup for easy assembly/disassembly, which also decreases heat transfer. This has worked well for class A amps during prototyping too. My way to see that there are some margins for long time use
So, some weak spots compared to Paroxod4's amp, but it's still a bit strange, since I'm running only 30V rails.
I did not check for oscillation, but I doubt it would be the cause since there is no feedback around the output..?
They are however NJW, not MJL as I mentioned before, so smaller package and lower heat dissipation, lower Cob, but very similar other than that. I also, mounted them with sil-pads in the proto-setup for easy assembly/disassembly, which also decreases heat transfer. This has worked well for class A amps during prototyping too. My way to see that there are some margins for long time use So, some weak spots compared to Paroxod4's amp, but it's still a bit strange, since I'm running only 30V rails.
I did not check for oscillation, but I doubt it would be the cause since there is no feedback around the output..?