Here's the corrected transistor setup. No BJTs or Mosfets were harmed during the process! Hehe
I saved all devices and tested them, also checked all resistors around to ensure none got damaged. screws and nuts are just there to align everything when soldering.
View attachment 637794
View attachment 637795
Thanks for letting me know Bigun! Good thing I posted some pictures!
Do
Hardly wait your feedback on these amplifier. I know you built a lot of amps has a lot of experience- I followed some of your projects. I have a set of PC boards also. 😀
Member
Joined 2009
Paid Member
What about C3, do I have the correct orientation? I believe I have to reverse it, after reading a post, seem to suggest it was wrong on the schematic.
C3 should be installed with +ve to ground (the schematic in Post #604 is wrong for C3 orientation). I had it the wrong way around when I built it and nothing bad happened but it's not good practice.
Member
Joined 2009
Paid Member
will have to adjust the compensation anyway for a low gain version like headphone amp.
The Singleton Input Stage looks really interesting to me.
Z
Hi Z,
The TGM8 is a high-feedback design, with two-pole compensation. Not sure if you need this approach when you are aiming for low gain ? but you could simulate it and see how it looks. I like the idea of the DB output, I have a pcb in hand (my TGM11) but not built (and it's open loop). I do recommend the Singleton input.
ok, board all checked and mounted on heatsink, no shorts anywhere. I just realized that my PSU is only +/-33Vdc so I'll get very little power out of the amplifier but enough to go through testing phase. I have a Cresnet SMPS of +/-42Vdc which could be a little better for initial testing. Some Toroidy transformers are supposed to show up next Thursday.
Do
Do
Member
Joined 2009
Paid Member
Hi Do,
Be careful of running at lower voltage - some parts values may not be optimal. For example, with D5 as a 30V zener and +/-33V supply the solid-state output relay may not be energized and you won't get anything on the output connector. The diode could be temporarily shorted out - the consequence will be that the output relay will not close as the power rails collapse but no harm should result.
Be careful of running at lower voltage - some parts values may not be optimal. For example, with D5 as a 30V zener and +/-33V supply the solid-state output relay may not be energized and you won't get anything on the output connector. The diode could be temporarily shorted out - the consequence will be that the output relay will not close as the power rails collapse but no harm should result.
Well... The amp measured very good and was playing really good on the suicide speaker but I think I might have pushed it a little too much because the speaker protection kicked in and won't go anymore... Upon inspection, seems like one of the isolation pad had a tiny metal particle that punctured it since all output transistors were making connection to gnd from center pin... Replaced the pads and fixed that issue but seems like the speaker protection SSRs might be dead? I'm playing with offset and measuring at the same time and the offset is everywhere... Meaning if I turn a little clockwise or counter-clockwise it goes crazy high, there does not seem to be a way to adjust anymore. Even with 4 volts on output offset, the led goes red but I can still measure the offset. Something busted I'm sure. No magic smoke anywhere though. Output transistors can still bias properly it seems... I'm measuring 50mV across R30 and R31
Wish I had not done this test... Won't assemble second channel now since I won't have a working pair.
Anyone has a clue to test something?
Thanks
Do
Wish I had not done this test... Won't assemble second channel now since I won't have a working pair.
Anyone has a clue to test something?
Thanks
Do
Member
Joined 2009
Paid Member
Ouch! Well, things happen. I don't knw if I fully understand your description of events but it seems you were pushing it hard and it failed, speaker protection kicked in. So you were playing loud music - that shouldn't have been a problem unless you were torturing it into heavy clipping.
Did you find a metal particle short or you guess it was there ? a short like this essentially grounds the amplifier output to the heatsink and excessive current would flow through the protection devices and the output devices and if excess base current was induced through the base junction of the output device it could stress other circuit elements.
From your description, the protection FET's failed short circuit and will have to be replaced.
It's difficult to diagnose what else is damaged from these symptoms but I'll make some further observations....
1) If the output is 4V then the output is not shorted to ground - yes you fixed that short. It also means that the output is not shorted to one of the power rails, so we also do not have a power device failed to a direct short.
2) it is interesting that the dc offset is close to the turn-on voltage of a power FET.
I will think and come back to you with a possible scenario.
Did you find a metal particle short or you guess it was there ? a short like this essentially grounds the amplifier output to the heatsink and excessive current would flow through the protection devices and the output devices and if excess base current was induced through the base junction of the output device it could stress other circuit elements.
From your description, the protection FET's failed short circuit and will have to be replaced.
It's difficult to diagnose what else is damaged from these symptoms but I'll make some further observations....
1) If the output is 4V then the output is not shorted to ground - yes you fixed that short. It also means that the output is not shorted to one of the power rails, so we also do not have a power device failed to a direct short.
2) it is interesting that the dc offset is close to the turn-on voltage of a power FET.
I will think and come back to you with a possible scenario.
Last edited:
Member
Joined 2009
Paid Member
Out of interest: With the power off, measure the conductance across the solid state relay (back to back FETs) and see if you can confirm if it failed short, or to a low resistance instead of the correct value of open circuit ?
Member
Joined 2009
Paid Member
The reason I ask about the protection relay is that the short to the heatsink would not stress the output relay so why would it fail - the offset appearing at the output of the relay is strange, but if the output is floating then it's voltage is ill-defined. The conductivity test will help understand.
Hi Bigun,
I will be able to test on Monday or Tuesday and will let you know.
I did not find any particles through the pads but there was continuity between the output devices and heatsink so I'm assuming something was making contact, which was definitely not at first when I set it up for test. Changing the pads fixed that issue but I'm assuming this stressed some parts from the amplifier.
So I will test conductance and let you know.
This is kinda sad since it sounded really good and bass was very well controlled.
Thanks
Do
I will be able to test on Monday or Tuesday and will let you know.
I did not find any particles through the pads but there was continuity between the output devices and heatsink so I'm assuming something was making contact, which was definitely not at first when I set it up for test. Changing the pads fixed that issue but I'm assuming this stressed some parts from the amplifier.
So I will test conductance and let you know.
This is kinda sad since it sounded really good and bass was very well controlled.
Thanks
Do
Member
Joined 2009
Paid Member
well don't get too upset, the journey and the challenge will be worth it. worse case, you replace active parts and it'll be fine, it would just be nice to avoid swapping out too many parts but I noticed your soldering skills are very good so it's all doable.
I was awake at 4am this morning thinking what might be the problem but even with a Spice simulation I couldn't figure it out so I went back to sleep.
You could build up the other channel, then compare voltages at various points in the circuit between good and bad channel to help track down what needs to be fixed
I was awake at 4am this morning thinking what might be the problem but even with a Spice simulation I couldn't figure it out so I went back to sleep.
You could build up the other channel, then compare voltages at various points in the circuit between good and bad channel to help track down what needs to be fixed
Hi Bigun,
Sorry to have kept you awake at such an early morning time. I will check the SSR MOSFETS and will also build the second channel as per your suggestion
Thanks !
Do
Sorry to have kept you awake at such an early morning time. I will check the SSR MOSFETS and will also build the second channel as per your suggestion
Thanks !
Do
I've removed the SSR Mosfets and bypassed with a jumper. Measured about 30mV accross R30 and R31 and tried as much as I could to lower the offet. It drifts a lot. If I set it to 0mV it will easily drift to 70mV a minute after. I connected my suicide speaker and source and the amplifier seems to be playing fine...
I'm about to remove the Output Mosfets and test with just BJTs. What are your thoughts?
Thanks
Do
I'm about to remove the Output Mosfets and test with just BJTs. What are your thoughts?
Thanks
Do
Last edited:
Member
Joined 2009
Paid Member
Good progress !
The amp should work fine without the MOSFETs but don't run it hard as you'll only have one pair of outputs for total current and the performance into difficult loads or on heavy bass may be worse without them.
I'm not sure why the offset is moving around, but one step at a time. Test the parts.
The amp should work fine without the MOSFETs but don't run it hard as you'll only have one pair of outputs for total current and the performance into difficult loads or on heavy bass may be worse without them.
I'm not sure why the offset is moving around, but one step at a time. Test the parts.
So I've removed the Mosfets and everything is still the same. Music is playing but offset drifts a lot. If i'm very patient, I can set it to almost 0 then it drifts up to 130mV and then goes down again to -50mV.
I've replaced Q1 and Q2 but the parts tested proper.
Do you think that the output BJTs could potentially be damaged but still play music and this causing the drifting?
I just don't know what else to look for. Maybe the drivers got damaged..?
I'm lost...
Do
I've replaced Q1 and Q2 but the parts tested proper.
Do you think that the output BJTs could potentially be damaged but still play music and this causing the drifting?
I just don't know what else to look for. Maybe the drivers got damaged..?
I'm lost...
Do
Member
Joined 2009
Paid Member
It is awhile since I tested the amplifier myself, you make me wonder how stable the offset was for me. My recollection is that it was very stable. The dc-offset, once set, is controlled by the negative feedback loop which has maximum control at 0Hz due to C3. The dc-offset is set by the voltage on C3 that is injected via R12. The dc-offset circuit that determines this voltage includes the zener diode D1 and the transistor Q2. If we have a drift in voltages in that circuit it can affect the dc-offset at the output. Can you measure the voltage at the Zener diode to see if it is stable (it may not be exactly -12V) ? Also look at the voltage at the emitter of Q2. Both voltages should be stable, if they are not this may be the cause of the drift.
The bias current was also stable - since the temperature feedback loop is between one of the driver transistors and the Vbe transistor. They are mounted together on the pcb, are they in good contact, I used thermal grease or a small isolation pad and then I bolted them together. The drivers do control the current through the output devices in this CFP architecture - do you have stable bias ?
The bias current was also stable - since the temperature feedback loop is between one of the driver transistors and the Vbe transistor. They are mounted together on the pcb, are they in good contact, I used thermal grease or a small isolation pad and then I bolted them together. The drivers do control the current through the output devices in this CFP architecture - do you have stable bias ?
Bias is very stable and does not drift once it reach its set point. DC offset was also very very stable when everything was working properly before the incident. I will check the voltages and let you know.
Thanks
Do
Thanks
Do
Member
Joined 2009
Paid Member
Another couple of thoughts.
There is a start-up period needed for things to settle. I had designed the speaker protection circuit to also provide a delay before connecting the speakers until the start-up period was completed. Many amplifiers can cause a noise or movement of cone in the speaker when the amplifier is starting up and a delay is commonly used to address that. The start-up period is not very long, more of a long pulse than a slow drift so this may not be the issue. With the protection FETs bypassed you don't have the start-up speaker delay feature enabled at present so the start-up will be measured at the ouput.
Secondly, as a Singleton input device, it will have some thermal drift. The device Q2 should track temperature with Q1, just the environment within the chassis, so there is a thermal feedback there. If you have the amp on the bench with a flow of air over it then this might affect the stability of the offset.
Lastly, perhaps Q1 was damaged slightly from the earlier experiments.
There is a start-up period needed for things to settle. I had designed the speaker protection circuit to also provide a delay before connecting the speakers until the start-up period was completed. Many amplifiers can cause a noise or movement of cone in the speaker when the amplifier is starting up and a delay is commonly used to address that. The start-up period is not very long, more of a long pulse than a slow drift so this may not be the issue. With the protection FETs bypassed you don't have the start-up speaker delay feature enabled at present so the start-up will be measured at the ouput.
Secondly, as a Singleton input device, it will have some thermal drift. The device Q2 should track temperature with Q1, just the environment within the chassis, so there is a thermal feedback there. If you have the amp on the bench with a flow of air over it then this might affect the stability of the offset.
Lastly, perhaps Q1 was damaged slightly from the earlier experiments.
Last edited:
I've replaced Q2 and Q1 already but no different. The amplifier is only on a heatsink for now and it is in the same environment as initially tested. Even without the chassis, it was pretty stable offset. We're talking of over 150mV of drift and this in the spam of a minute or so. It is definitely not showing the same behavior as before, so something if defect somewhere. I'll do the checkup voltage on Q2 and D1.
Thanks
Do
Thanks
Do
- Home
- Amplifiers
- Solid State
- TGM8 - my best amplifier, incredible bass, clear highs, no fatigue (inspired by Rod Elliot P3a)