Current measurement of Left channel of this F5T is THD+noise < 0.00126. -78 dbV, THD alone -88 dbV. The second harmonic is the largest spike.
Noise on the chart never exceeds -90 dB.
Input signal is -23dBV, peak is -0.15 dBV. Into an 8 ohm load -- so about 1 watt.
Woot!
Three or so changes left to make, cause the wiring is not all properly done. Should be even quieter.....
Noise on the chart never exceeds -90 dB.
Input signal is -23dBV, peak is -0.15 dBV. Into an 8 ohm load -- so about 1 watt.
Woot!
Three or so changes left to make, cause the wiring is not all properly done. Should be even quieter.....
Right channel not so quiet.... there is a small toroid that runs the ancilliary circuits, like speaker protection. The zero volt like crosses it. The noise is at 60Hz and multiples. This is not present on the left channel. -68dB THD+Noise, THD is -82dB. Right channel at 1 watt output.
Needs a fix. Perhaps spare fifth winding to power this, keeping all noise sources at one end of the amp....
Needs a fix. Perhaps spare fifth winding to power this, keeping all noise sources at one end of the amp....
Moving the aux transformer has zero effect. I rewired the FE board to the ground loop breaker with a thicker wire, and it got worse. So it seems that the noise on the 0 volt line is being fed back to the input stage and modulating Vgate.
So, I tried wiring RCA Shield to link and then link to GND with a 1 ohm resistor.
THE 60 Hz HUM DROPPED TO -88 DB
That is nearly a 20 dB drop. It was at -69.5 dB.
So, I tried wiring RCA Shield to link and then link to GND with a 1 ohm resistor.
THE 60 Hz HUM DROPPED TO -88 DB
That is nearly a 20 dB drop. It was at -69.5 dB.
Same on the other channel, the 60 Hz dropped to -91.4 dB.
However, on this channel, there is 120 Hz at -78 dB. It might be that the rectification for the ancilliary power supply is too near the power feed for this FE board. Will try to move this 24 volt rectification elsewhere.
Does any one build amps with 12 VDC wallwarts to drive the ancilliary/secondary power? Seems this would be an excellent way to keep a noise source out of the box.
However, on this channel, there is 120 Hz at -78 dB. It might be that the rectification for the ancilliary power supply is too near the power feed for this FE board. Will try to move this 24 volt rectification elsewhere.
Does any one build amps with 12 VDC wallwarts to drive the ancilliary/secondary power? Seems this would be an excellent way to keep a noise source out of the box.
It looks like I have a bit of trouble...
That issue with the different DC offset vs RCA shorting plug installed is back. Shows 250 mV DC offset without RCA shorting plug, and 0 mv with RCA shorting plug in place.
Possibly bad contact from feedback trace to link?
Boards have been heated a lot....
Could this be bad P3? But, thd in affected channel is vanishingly low... 0.0012%
That issue with the different DC offset vs RCA shorting plug installed is back. Shows 250 mV DC offset without RCA shorting plug, and 0 mv with RCA shorting plug in place.
Possibly bad contact from feedback trace to link?
Boards have been heated a lot....
Could this be bad P3? But, thd in affected channel is vanishingly low... 0.0012%
Last edited:
Yes. I am thinking an impedance problem... Thinking about replacing P3. The THD from one channel to the other is much different -- the other side has -78 dB of 2nd order, and this side is down around -90. This channel is Fairchild, the other is IRF240/9240
I have a scope.
Thanks.
I have a scope.
Thanks.
Last edited:
I did not want to cut any traces with this modification.
But, the 47.5k R2, is taken directly to ground by the board, not to the link pin, making the shield and the FE amp ground at a slightly different potential. Is this enough?
Disconnect the ground leg of R2 and wire a jumper to link. (actually to the link side of R4 -- spare contact open. )
R shield to case = 15 ohms whether or not shorting pin is in place. Linke side of R4 to Gnd 1.6 Ohms.... ( 0.6 is DMM leads ).
DC offset,
Shorting pin IN -35mV
Shorting pin OUT +300 mV
Even worse this way..... this was there before the HBR mode.
Now, I am thinking oscillation.
But, the 47.5k R2, is taken directly to ground by the board, not to the link pin, making the shield and the FE amp ground at a slightly different potential. Is this enough?
Disconnect the ground leg of R2 and wire a jumper to link. (actually to the link side of R4 -- spare contact open. )
R shield to case = 15 ohms whether or not shorting pin is in place. Linke side of R4 to Gnd 1.6 Ohms.... ( 0.6 is DMM leads ).
DC offset,
Shorting pin IN -35mV
Shorting pin OUT +300 mV
Even worse this way..... this was there before the HBR mode.
Now, I am thinking oscillation.
Last edited:
Oh, one thing I should mention about this implementation.
The channel in question has MOSfets from two different batches.
The boards are in parallel: power, gate, output and ground are all independently wired to each board.
I should move the FE board from the to see if the problem goes with the board. Then I might know where to look.
The channel in question has MOSfets from two different batches.
The boards are in parallel: power, gate, output and ground are all independently wired to each board.
I should move the FE board from the to see if the problem goes with the board. Then I might know where to look.
Last edited:
Scope readings:
Frequency:
with plugs in: 25 MHz rarely jumping to 500 Mhz -- maybe me moving
with plugs out: 25MHz jumping to 500 MHz a lot, sometimes showing 1GHz.
Vavg around 1mV.
Regardless, this is above the bandwidth of the amp, possibly catching noise from water meter transmitter/cell phone.
Therefore, amp oscillation does not exist.
Frequency:
with plugs in: 25 MHz rarely jumping to 500 Mhz -- maybe me moving
with plugs out: 25MHz jumping to 500 MHz a lot, sometimes showing 1GHz.
Vavg around 1mV.
Regardless, this is above the bandwidth of the amp, possibly catching noise from water meter transmitter/cell phone.
Therefore, amp oscillation does not exist.
Last edited:
The resistance from output to ground on the FE board is 65.7 ohms. This is correct. The resistance on the good board is 65.9 ohms. All resistors are soldered properly.
When the shorting plug is added, the resistance from the gates of the Jfets to ground changes. It goes from 48.5K to 1K. If the soldering job is lousy at the jfets, different voltages will appear a the gates when the impedance to ground changes.
I think reflowing the jfet connections is mandatory. If that fails, then possibly replacing the jfets.... with all the damage this amp has seen, they may have been stressed and no longer match.
This is the most sensible reasoning yet, as the only impedance change with rca in or out is in reference to the jfets.
When the shorting plug is added, the resistance from the gates of the Jfets to ground changes. It goes from 48.5K to 1K. If the soldering job is lousy at the jfets, different voltages will appear a the gates when the impedance to ground changes.
I think reflowing the jfet connections is mandatory. If that fails, then possibly replacing the jfets.... with all the damage this amp has seen, they may have been stressed and no longer match.
This is the most sensible reasoning yet, as the only impedance change with rca in or out is in reference to the jfets.
I think this amp is possessed.
Jfets changed to a fresh pair .. these are marked 7.3 Idss.
No difference... except maybe now, the bias does not get high enough.
Two meters on output boards on the bad channel. Biasing done with the shorting plug out. Readings across the source resistors are:
Shorting plug out, P-Channel 263 mV, N-channel 238 mV, DC offset 2 mV.
Shorting plug in, P-Channel 263 mV, N-channel 238 mV, DC offset -0.31 V
??? WHAT ???
These two Mosfets have the same current going through them in both cases, yet, the DC offset is wildly different.
The DC offset bounces around quite a bit in both cases, it moves around +/-10 mV in 10 seconds. This was never so unstable.
Ohming out FE output to gate show 1.063K on both channels.
Jfets changed to a fresh pair .. these are marked 7.3 Idss.
No difference... except maybe now, the bias does not get high enough.
Two meters on output boards on the bad channel. Biasing done with the shorting plug out. Readings across the source resistors are:
Shorting plug out, P-Channel 263 mV, N-channel 238 mV, DC offset 2 mV.
Shorting plug in, P-Channel 263 mV, N-channel 238 mV, DC offset -0.31 V
??? WHAT ???
These two Mosfets have the same current going through them in both cases, yet, the DC offset is wildly different.
The DC offset bounces around quite a bit in both cases, it moves around +/-10 mV in 10 seconds. This was never so unstable.
Ohming out FE output to gate show 1.063K on both channels.
Last edited:
With nothing:
Voltage across one of the P-channel sources was measured at 0.45 V, another at .16 and another at .260, jumping to 0.375. This channel measured extremely well for THD... could it be because one MOSfet was running so hot?
Going to install a fully matched quad.
=====
6L6 Just saw your message. I will plug in the source tomorrow..... been at it a while tonight and don't want to make mistakes.... wife is out of town, so good time to do this. A fully matched set is the right thing to put in now -- need to make sure nothing blows up.
THANKS
Voltage across one of the P-channel sources was measured at 0.45 V, another at .16 and another at .260, jumping to 0.375. This channel measured extremely well for THD... could it be because one MOSfet was running so hot?
Going to install a fully matched quad.
=====
6L6 Just saw your message. I will plug in the source tomorrow..... been at it a while tonight and don't want to make mistakes.... wife is out of town, so good time to do this. A fully matched set is the right thing to put in now -- need to make sure nothing blows up.
THANKS
Last edited: