because the inherent noise that voltage regulators form on their ground. r18 is B+ ground and chassis ground which can cause common mode noise to develop at the input of the unbalanced I/O circuit.
with tubes removed and power applied, look at the heater 12V and 0V with the oscilloscope connected to signal ground. what you should see is a strait line. If the heaters were AC, it should be a clean sine wave. with the tubes plugged in and high voltage B+ and bias diodes removed, there should be a strait line when measuring the cathode, with the heater power only applied to the tubes.
with tubes removed and power applied, look at the heater 12V and 0V with the oscilloscope connected to signal ground. what you should see is a strait line. If the heaters were AC, it should be a clean sine wave. with the tubes plugged in and high voltage B+ and bias diodes removed, there should be a strait line when measuring the cathode, with the heater power only applied to the tubes.
I can't remember the exact value, maybe 1k. i basically set it to reduce the brightness on the Led.
maybe i could use the zener string to regulate the B+, and get the voltage for the led from somewhere else.
@Davesnothere : Sorry R18 doesnt exist. thats leftover from the original circuit. i have only one connection from signal earth to chassis earth, and thats R1 on the preampcircuit.
the regs are not connected to the chassis at all, the only semconductors connected to the chassis are the regs (7812) and the ccs for the driver stage.
They have this silicon thingy inbetween and i used nylon as well.
seeing that my amp has nylons already maybe it just needs a bit of latex and it will be fixed......
ahm if anything sounds weird, feel free to ask , i am not a native speaker, well, not english that is.
Last edited:
Hello Mel, thanks for the reminder.
i haven't, mostly because they are soldered in and a bit of a pain to remove where i put them, and i remember calculating them with f= 1Hz,
rk' =rcathode//rk = 682Ohm
Ck=1/(6.28*f*rk')= 273uF
i chose next size i had, 470uF (seeing that there are lots of designs with 1000uF),
and that gives me f=0.5Hz... so ithought that whould be ok.
Hello DF96, i imagined it's not as easy ...
i ordered some parts so i can try some of the suggestions, i dont expect them to be here before friday.
i am testing the right channel of the amp. the left channel is connected to the supply, with a load and with the input grounded. I am wondering if it would be best to disconnect left, so it can't interfere with the one i am testing? is it enough to just disconnect the b+ rail from the CCS V1/V2?
some testing done,
first i removed B+ from the left channel to make sure i only measure whats happening on the right. i conected the pc to the input terminal, feeding 40Hz 0.2V. NFB loop is open, and increase the volume.
-i removed R13 from the loop, LED is off, but thats the only change. the bias balance circuit (attached) reacts, i see the LED flashing at LF.
-as suggested by MelB i decreased C23, first to 250uF and then to 150uF with no change. same as above.
looks like i have to wait for a few zeners and caps to try some other suggestions
first i removed B+ from the left channel to make sure i only measure whats happening on the right. i conected the pc to the input terminal, feeding 40Hz 0.2V. NFB loop is open, and increase the volume.
-i removed R13 from the loop, LED is off, but thats the only change. the bias balance circuit (attached) reacts, i see the LED flashing at LF.
-as suggested by MelB i decreased C23, first to 250uF and then to 150uF with no change. same as above.
looks like i have to wait for a few zeners and caps to try some other suggestions
Attachments
Here's what I think is going on:
I think the reason you're having issues with motor boating is that the "reference" for your CCS is derived by a resistive divider from the power supply. Take the "CCS" on the input pair: The "reference" is formed by R30, R31. I use the quotes (" ") to indicate that it really isn't much of a reference, it's just a fraction of the 320 V supply. The 320 V isn't exactly 320 V either. It's whatever B+ happens to be minus the voltage drop across R7+R8 in the supply. On start-up, there is very little drop across R7+R8. As the input tube starts conducting, the supply voltage to the input stage will drop as the voltage across R7+R8 increases. This causes the voltage across R30 to decrease. This causes the current of the "CCS" to decrease. This causes the supply voltage to increase (less drop across R7+R8). Now the voltage across R30 increases, hence, increasing the current in the "CCS". This is motor boating. I bet the frequency is roughly 1/(2pi*R30*C25). If you double C25, I bet the motor boating frequency will be half.
Similarly, the stability of the "CCS" on the phase splitter is dependent on the stability of the 24 V bias voltage.
The motor boating is caused by the current of the "CCS" being strongly dependent on the power supply voltage. Replacing R30 with a zener diode would be one way to fix this issue. It would result in a somewhat temperature dependent CCS current but at least the motor boating issue should be fixed. With the current circuit values, I suspect a 47 V zener would work. Or use one of the 51 V zeners in series with the "ON" LED for a quick test... This test would reveal if it's worthwhile building a different CCS or if the motor boating is caused by something else.
~Tom
I think the reason you're having issues with motor boating is that the "reference" for your CCS is derived by a resistive divider from the power supply. Take the "CCS" on the input pair: The "reference" is formed by R30, R31. I use the quotes (" ") to indicate that it really isn't much of a reference, it's just a fraction of the 320 V supply. The 320 V isn't exactly 320 V either. It's whatever B+ happens to be minus the voltage drop across R7+R8 in the supply. On start-up, there is very little drop across R7+R8. As the input tube starts conducting, the supply voltage to the input stage will drop as the voltage across R7+R8 increases. This causes the voltage across R30 to decrease. This causes the current of the "CCS" to decrease. This causes the supply voltage to increase (less drop across R7+R8). Now the voltage across R30 increases, hence, increasing the current in the "CCS". This is motor boating. I bet the frequency is roughly 1/(2pi*R30*C25). If you double C25, I bet the motor boating frequency will be half.
Similarly, the stability of the "CCS" on the phase splitter is dependent on the stability of the 24 V bias voltage.
The motor boating is caused by the current of the "CCS" being strongly dependent on the power supply voltage. Replacing R30 with a zener diode would be one way to fix this issue. It would result in a somewhat temperature dependent CCS current but at least the motor boating issue should be fixed. With the current circuit values, I suspect a 47 V zener would work. Or use one of the 51 V zeners in series with the "ON" LED for a quick test... This test would reveal if it's worthwhile building a different CCS or if the motor boating is caused by something else.
~Tom
OP said the motor boating was there even with the feedback disconnected, so I doubt it's a stability issue within the main feedback loop.
You're right about the 90 degrees phase shift of an RC, however. I just ran a sim of the input stage. My sim doesn't show signs of motor boating... Perhaps my thinking of earlier was incorrect. I've been proven wrong before...
I just happened to notice that just about every supply in the amp comes from a voltage doubler -- including the main B+ supply. I suggest monitoring all the supply voltages on an o'scope. I'm a little skeptical about pulling hundreds of mA from a voltage doubler, but I haven't worked that much with those. Of course, if all supplies are solid, the issue isn't with the supplies.
~Tom
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.