Last night I removed the EF86 circuit and went directly into the ECC83. Treble rolls off to -15dB at 20kHz. I was certain that it was all happening in the EF86, the roll off to -25dB was present at the EF86 cathode according to my probe! So I then removed the ECC83 circuits and put the EF86 back. Same, -15dB at 20kHz. I also rebuilt most of the power supply to make sure there were no bad solder joints. Starting to think it could be a grounding issue?
DF, my probing may not have been very accurate. It's a DIY jobbie. Just a 100n cap in an enclosure with a jack socket, a crocodile lead for ground and a probe. That plugs into a Palmer DI box (1M impedance) and then a mic input on my Tascam US-2000 audio interface. To measure the frequency response I use Waves FFT plugin, and a couple of free ones too. They all give slightly different results, but generally say the same thing.
For a signal source I'm using a VST plugin from mda doing a log sweep from 20Hz to 20kHz. I've checked that it does produce a flat output, also I've plugged my line output straight into a line in to check my interfaces frequency response, and it's almost flat, just a little ripple.
Originaly the line out of my audio interface went into a Palmer DI box then into my preamps XLR in and transformer. I've tried different line outputs, DI boxes and transformer secondary loads with no joy so I eliminated all of that and now plug the line out straight into the grid with a 1M grid leak. (I've also tried the headphone output to grid).
Most tests are from the preamps output into a 10kohm line in on my interface, as it would be used in the real world. I think these crude tests are good enough for now, It's not a phantom problem from bad test methods, an SM57 sounds dull and muffled as does my valve condenser mic.
I've checked my circuitry and it IS what is in the schematic. BTW the 1M log pot in the schematic is not a gain control, it's an output control. I havent had a gain control since I made it all DC coupled. A 100k dual gang pot was going to vary the cathode bypass on the ECC83s.
DF, my probing may not have been very accurate. It's a DIY jobbie. Just a 100n cap in an enclosure with a jack socket, a crocodile lead for ground and a probe. That plugs into a Palmer DI box (1M impedance) and then a mic input on my Tascam US-2000 audio interface. To measure the frequency response I use Waves FFT plugin, and a couple of free ones too. They all give slightly different results, but generally say the same thing.
For a signal source I'm using a VST plugin from mda doing a log sweep from 20Hz to 20kHz. I've checked that it does produce a flat output, also I've plugged my line output straight into a line in to check my interfaces frequency response, and it's almost flat, just a little ripple.
Originaly the line out of my audio interface went into a Palmer DI box then into my preamps XLR in and transformer. I've tried different line outputs, DI boxes and transformer secondary loads with no joy so I eliminated all of that and now plug the line out straight into the grid with a 1M grid leak. (I've also tried the headphone output to grid).
Most tests are from the preamps output into a 10kohm line in on my interface, as it would be used in the real world. I think these crude tests are good enough for now, It's not a phantom problem from bad test methods, an SM57 sounds dull and muffled as does my valve condenser mic.
I've checked my circuitry and it IS what is in the schematic. BTW the 1M log pot in the schematic is not a gain control, it's an output control. I havent had a gain control since I made it all DC coupled. A 100k dual gang pot was going to vary the cathode bypass on the ECC83s.
Last edited:
DF, I hope you're onto something there. Though I have tried a 6 inch speaker lead from the headphone output into the grid/1M. And a 10M guitar lead from the line output into the grid/1M. I assumed that the long lead wouldn't be the problem as this is what I used check the interfaces own frequency response, and found it to be flat(ish).
I'll see what cables I can find/make. I'll be back soon.
I'll see what cables I can find/make. I'll be back soon.
I hope I understand you correctly. You are using a mic-in on your sound card to measure with. Could it be you are overdriving the input and get inaccurate results due to massive distortion?
From what you write I guess that the problem is not in your circuit but in the way you measure. You have the same roll off in the EF86 stage and in the ECC83 stage. Highly unlikely!
Get in contact with a friend who has an o-scope!!
From what you write I guess that the problem is not in your circuit but in the way you measure. You have the same roll off in the EF86 stage and in the ECC83 stage. Highly unlikely!
Get in contact with a friend who has an o-scope!!
grommeteer, my sound-card has 8 mic inputs and 8 line inputs. I've just done the same measurements on an SPL Gold Mike preamp and it seems pretty flat from 40Hz to 18kHz. My methods are pretty naff, but I think good enough for my purposes. I wish I had a friend with an o-scope!
When necessary I use the -30dB pad on the DI box with the audio probe, so I never distort my sound-cards mic input.
DF, Also, I've just used a 1m jack cable to get the sine sweep into my preamp, still the same roll off. This evening I'm going to simplify the grounding. Bye bye star ground.
When necessary I use the -30dB pad on the DI box with the audio probe, so I never distort my sound-cards mic input.
DF, Also, I've just used a 1m jack cable to get the sine sweep into my preamp, still the same roll off. This evening I'm going to simplify the grounding. Bye bye star ground.
Last edited:
It's great that people on this forum could be that helpful (and trusting), but I think I can manage with what I've got (for this project anyway!).
Well, I've re-organised the grounding and the only change is a shedload more noise.
Going to keep going though, there's a chance I may not have sanded off enough powder coat to make good contact with the chassis.
Could it maybe be the quality of my filter caps?
Well, I've re-organised the grounding and the only change is a shedload more noise.
Going to keep going though, there's a chance I may not have sanded off enough powder coat to make good contact with the chassis.
Could it maybe be the quality of my filter caps?
No, component quality does not make that sort of change.
As I said, either the circuit is not what you think you have built or you are not measuring what you think you are measuring. Circuits are causal; impossible things don't happen so there is a rational explanation. Remote debugging is not easy!
As I said, either the circuit is not what you think you have built or you are not measuring what you think you are measuring. Circuits are causal; impossible things don't happen so there is a rational explanation. Remote debugging is not easy!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.