@kevinkr no offense taken here. I thought maybe they were saying it was pointless to investigate if I were just going to filter out the noise at input stage (though I only moved that RC pole temporarily to see how it would impact things). Even if it was meant as amore negative, comment it's all good. This stuff can be frustrating and I appreciate all the advice and appreciate your diligent moderation.
I began looking into those leads yesterday and I really appreciate your help. I was looking at audio vs plate voltage on my scope yesterday, but need a bit more time to more carefully observe and analyze. Preliminary thoughts: there does seem to be correlation between plate voltage jitter and audio peaks. Plate voltage is relatively stable and waivers up to +/- 0.05V if I'm watching in AC mode. Sometimes there are momentary spikes that waiver a bit more than that. I need to examine more carefully though. Thanks again for your help and patience.
Will do. I know you've asked this before and my apologies for not addressing directly yet. Per @craigtone comments, I cannot share his schematic though I can offer some general information that may be helpful.
In short, the heater/filament power supply is regulated (via LDO) 6.3VDC. I'm going from a 6.3VAC transformer secondary, to a full wave rectifier bridge (with 4 Schottky diodes), and to an LDO regulator circuit. Heaters on the 3 tubes are wired in parallel. Voltage looks rock solid on my oscilliscope and is calibrated for 6.3VDC on the heaters.
The HV side is fed from a 240VAC transformer secondary, goes through a full wave rectifier bridge, a parallel choke, and to a MOSFET transistor (IRF840). Plate voltage is around 246VDC if I recall correctly. Voltage is relatively stable though does jitter a bit if examining in high resolution. For example, plate can bounce up or down up to 0.05 Volts if observing in AC mode on the scope with occasional momentary spikes outside that range. As stated in #65, I need to examine this more carefully though. These results may not fall within the expected outcomes of @craigtone 's power supply. He may be able to clarify my description.
I'll completely understand if it's not possible to offer more precise guidance without more precise details. Your advice has been helpful.
I did not know this is a Top Secret Design.
Is the filament supply grounded in any way (if on pcb it certainly is or should be) ?
Is the filament supply grounded in any way (if on pcb it certainly is or should be) ?
Preamp circuit is very close to WAD preamp schematic in post #26 and I'm not sure about the power supply. PCB and final schematic was designed by @craigtone and I'm respecting his wishes to not post it (per post #39). I certainly understand that folks won't be able to pinpoint issues without more precise details, and yet folks have already gone above and beyond my expectations in providing helpful guidance. This is all a learning process for me.
Filament supply is not connected to "star ground" which goes to the chassis.
I'm certainly no expert , but I've built 4 line stage tube preamps .
Have also read through much of this - a few suggestions - from experience .
A phono stage has got something in the order of 32 dB's of gain - so its hypersensitive to anything .
- I have heard of situations when E core transformer physically transmits hum to microphonic tubes .
For this reason , I usually a bit of foam underneath boards and mount the transformer on rubber grommets .
In your case the phono board could be mounted on rubber gromments .
- Did you get the star ground node in place ?
- IMO all shielding should be connected to analog ground - NOT hydro earth .
- Total agree with the recommendation of adding an RC between the plate supply and board .
In my case it stopped HF oscillation with a line stage tube preamp .
- There can be a beautifully clean plate supply . Also recall the heaters should not be left floating wrt to the cathodes .
So if the heater supply is not lifted , the heater ground , is connected directly to the nice clean analog ground .
Then what was a nice clean analog ground is all messed up .
So try connecting the heater ground to the analog ground with something like 47K in between .
The voltage drop across the 47K will be next to nothing .
- As a last resort - if everything else has failed . The heater supply and plate supply are running off
the same toriod transformer . Now suppose the heater diodes switch off and cause ringing
which is then picked up by the secondary for the plate supply .
As a last resort try a separate transformer for the heater supply .
.
Have also read through much of this - a few suggestions - from experience .
A phono stage has got something in the order of 32 dB's of gain - so its hypersensitive to anything .
- I have heard of situations when E core transformer physically transmits hum to microphonic tubes .
For this reason , I usually a bit of foam underneath boards and mount the transformer on rubber grommets .
In your case the phono board could be mounted on rubber gromments .
- Did you get the star ground node in place ?
- IMO all shielding should be connected to analog ground - NOT hydro earth .
- Total agree with the recommendation of adding an RC between the plate supply and board .
In my case it stopped HF oscillation with a line stage tube preamp .
- There can be a beautifully clean plate supply . Also recall the heaters should not be left floating wrt to the cathodes .
So if the heater supply is not lifted , the heater ground , is connected directly to the nice clean analog ground .
Then what was a nice clean analog ground is all messed up .
So try connecting the heater ground to the analog ground with something like 47K in between .
The voltage drop across the 47K will be next to nothing .
- As a last resort - if everything else has failed . The heater supply and plate supply are running off
the same toriod transformer . Now suppose the heater diodes switch off and cause ringing
which is then picked up by the secondary for the plate supply .
As a last resort try a separate transformer for the heater supply .
.
Last edited:
Can you measure any resistance between filament and ground ?
But this may not be the (main) reason for your low freq noise.
But this may not be the (main) reason for your low freq noise.
About 2ohms. And I was mistaken about negative filament not tying back to my board ground. It does.
You wrote in post 67 "Voltage is relatively stable though does jitter a bit if examining in high resolution.
For example, plate can bounce up or down up to 0.05 Volts if observing in AC mode on the scope with
occasional momentary spikes outside that range .."
Correlation between lf output noise and hv supply noise can not be 100% due to the RIAA shaping and
maybe additional filtering which is probably not shown in the schematic of post 26. There is no useful
supply rejection in a circuit as seen there.
For example, plate can bounce up or down up to 0.05 Volts if observing in AC mode on the scope with
occasional momentary spikes outside that range .."
Correlation between lf output noise and hv supply noise can not be 100% due to the RIAA shaping and
maybe additional filtering which is probably not shown in the schematic of post 26. There is no useful
supply rejection in a circuit as seen there.
Hi, I encountered strange sounds in an E88CC and ECC88.
I used an SMPS to get 12V, and initially had it on earth. A lot of unspecified and intermittent noise.
Then I connected it to a 40V source, everything clear. This was a driver for an output stage amplifier 5x.
Im not sure it is of any help but the ‘bias’ of the filament power is not mentioned; maybe a black swan.
I made about the same type of RIAA two times. [with ECC88 cascode input]. Now I have EAR.
I used an SMPS to get 12V, and initially had it on earth. A lot of unspecified and intermittent noise.
Then I connected it to a 40V source, everything clear. This was a driver for an output stage amplifier 5x.
Im not sure it is of any help but the ‘bias’ of the filament power is not mentioned; maybe a black swan.
I made about the same type of RIAA two times. [with ECC88 cascode input]. Now I have EAR.
One thing I learned early by pure accident is that when using one toroid for HV/B+ & heaters - the secondaries should be as far away as possible - if they are very close or together capacitive coupling happens / ground loop - which results in a hum that can be heard and opposite - when removed the hum is significantly lower.
The graph for the original circuit, the scope video you showed earlier and the fact that you see variations on the supply all seem consistent with filtered subsonic noise coming from the supply (due to mains fluctuations for example). Above 5 Hz or so, the noise drops at just over 18 dB/octave. You could try increasing the filter time constants in the supply, or adding more filtering (as was already suggested).
The output signal of my phono amplifier also moves up and down randomly, but slower than yours. I guess I used larger power supply filter time constants.
The "pole moved to 10 Hz" plot looks like there is something motorboating at about 10 Hz. That is, there is a spectral bump around each multiple of 10 Hz.
Is the circuit with the MOSFET a series regulator or some capacitance multiplier thing? Is it in dropout or working as intended?
My guess is that the HV regulator is simply noisy. This combined with the finite PSRR of the 1st stage and the high gain at low frequencies results in LF noise at the audio output.
You might try to "fix" the gate voltage of the HV regulator FET by a suitable electrolytic capacitor. I would add a reverse biased protection diode between G an S in this case. Worth a try.
You might try to "fix" the gate voltage of the HV regulator FET by a suitable electrolytic capacitor. I would add a reverse biased protection diode between G an S in this case. Worth a try.
Sure, but every little thing in routing can sometimes add little help.
To be of some use I had a phono which did not have any coupling between the stages and worked normal ( just tube hiss ) - but that one had very heavy passive filtering of 4x 470uF in parallel ( EI core power trafo ) and this solution was slightly better than multiple C-RC cells ( around 4 ).
Last edited: