Yesterday I finished making changes to my all-differential PP amp. One of the changes involved increasing the grid stopper on the input (non-grounded) grid of the 6SL7 LTP splitter, from 10k to 200k, to take advantage of the Miller effect and limit the high-frequency response. The intention is to have -3dB at 40kHz.
Another change was to add 1Meg local feedback resistors from the 6SN7 driver plates to the splitter plates and to reduce global feedback by 50%.
The ovedrall gain is now a bit higher but 50Hz hum has become audible, whereas before it wasn't. I haven't changed the layout significantly. My power transformers are toroids and shouldn't be radiating significant hum, AFAIK. I wonder if giving the IP tube a big stopper may have increased its senstivity fo hum? Or maybe I should have increased the stopper on the grounded grid too, to preserve PSRR? Any ideas?
BTW, there is no connection to ground at all, because I don't have a 3-core power cable. That sounds pathetic, I realize, and I intend to remedy it but, up until now, it hasn't made any audible difference.
Another change was to add 1Meg local feedback resistors from the 6SN7 driver plates to the splitter plates and to reduce global feedback by 50%.
The ovedrall gain is now a bit higher but 50Hz hum has become audible, whereas before it wasn't. I haven't changed the layout significantly. My power transformers are toroids and shouldn't be radiating significant hum, AFAIK. I wonder if giving the IP tube a big stopper may have increased its senstivity fo hum? Or maybe I should have increased the stopper on the grounded grid too, to preserve PSRR? Any ideas?
BTW, there is no connection to ground at all, because I don't have a 3-core power cable. That sounds pathetic, I realize, and I intend to remedy it but, up until now, it hasn't made any audible difference.
ray_moth said:
...Ouch? I'd think you'd hear that. Especially with something high-gain like SL.
Hey! Uh...WTF? There's no capacitance spec on the GE white sheet! 😕
Ok... RFT says 1.5pF, pretty low.
Ok. Then grid oscillation was unlikely the problem before, nor is it now.
Certainly, the excessive impedance at that location will allow much greater hum induction. I dare you to bring your finger near the grid wire and listen to the hum grow without touching it.
I don't know who gave you the idea that's a bad thing. Grounded cables invite ground loop trouble.
Your problem probably lies in PSU filtering or heater treatment, unless there happens to be a particularly suceptible connection picking it up.
Tim
Ray,
I replied to your duplicate post over at AudioAsylum.
Duplicating reply here for possible interest of others.
The problem is the increased grid stopper. Put it back the way it was and use a small cap anode to anode in the diff amp if you realy want to roll off the tops a bit.
As I pointed out in the Asylum post there is a good story in Allen Wright's Preamp Cookbook about just this problem except in his case the grid was being driven from a volume pot and the problem only manifested itself at certain volume settings (about half volume) - the settings which gave the largest (effective)resistance from grid back to 0V. Its due to the small residual grid currents and heater cathode / cathode grid leakages reacting with the larger resistor to produce larger noise voltage. The larger resistor will introduce larger thermal noise of its own as well.
Grid Stoppers are usually essential BUT never use more than you need and rolling off highs by using high grid stopper values to interact with the Miller capacitance is a guitar amp trick best left for use in guitar amps.
Are you really sure you want to roll off the HF this way. Its not really outside the global loop (assuming global feedback goes to the other side of the diff amp). If you really want to slug the HF find out where your dominant pole is and slug that else you are asking for stability problems - two poles too close to each other is a serious problem.
ASIDE: My Control Theory lecturer gave us a usefull "Rule of Thumb". The maximum gain you can use in a control loop is equal to the ratio of the 2 most dominant poles. This can be modified by step networks etc. but its still a useful "rule of thumb".
Also I'm not sure that the theory of rolling off the HF before the power amp to avoid TIM has actually been proven.
If you want to do that then a separate R and C at the input to produce the rolloff is a better option, using the Miller capacitance and the grid stopper to do it I reckon will screw the HF balance of the diff amp. The anode to anode cap avoids screwing the HF balance but may have stability issues if the pole is too close to the dominant pole.
Oh Yeh - and as stated in a post above , ahigher impeadnce circuit is more susceptible to noise pickup from adjacent circuitry, heater supplies, transformer magnetic fields etc.
Cheers,
Ian
I replied to your duplicate post over at AudioAsylum.
Duplicating reply here for possible interest of others.
The problem is the increased grid stopper. Put it back the way it was and use a small cap anode to anode in the diff amp if you realy want to roll off the tops a bit.
As I pointed out in the Asylum post there is a good story in Allen Wright's Preamp Cookbook about just this problem except in his case the grid was being driven from a volume pot and the problem only manifested itself at certain volume settings (about half volume) - the settings which gave the largest (effective)resistance from grid back to 0V. Its due to the small residual grid currents and heater cathode / cathode grid leakages reacting with the larger resistor to produce larger noise voltage. The larger resistor will introduce larger thermal noise of its own as well.
Grid Stoppers are usually essential BUT never use more than you need and rolling off highs by using high grid stopper values to interact with the Miller capacitance is a guitar amp trick best left for use in guitar amps.
Are you really sure you want to roll off the HF this way. Its not really outside the global loop (assuming global feedback goes to the other side of the diff amp). If you really want to slug the HF find out where your dominant pole is and slug that else you are asking for stability problems - two poles too close to each other is a serious problem.
ASIDE: My Control Theory lecturer gave us a usefull "Rule of Thumb". The maximum gain you can use in a control loop is equal to the ratio of the 2 most dominant poles. This can be modified by step networks etc. but its still a useful "rule of thumb".
Also I'm not sure that the theory of rolling off the HF before the power amp to avoid TIM has actually been proven.
If you want to do that then a separate R and C at the input to produce the rolloff is a better option, using the Miller capacitance and the grid stopper to do it I reckon will screw the HF balance of the diff amp. The anode to anode cap avoids screwing the HF balance but may have stability issues if the pole is too close to the dominant pole.
Oh Yeh - and as stated in a post above , ahigher impeadnce circuit is more susceptible to noise pickup from adjacent circuitry, heater supplies, transformer magnetic fields etc.
Cheers,
Ian
FWIW the same thing happens when the input cap to my high sensitivity amp (~100 mv for full out) is too small. The 10 meg input impedance notwithstanding, an increase from .02 uF to .25 uF substantially reduced the hum I thought was power supply induced.
Thanks, everyone! I reverted to 10k stoppers at both 6SL7 grids with no other HF limiting filters. I left the GFB resistor as it was because it seems to sound more open and detailed, less "muffled", than it was before, if you see what I mean. Anyway, the hum is gone and there is no sign of instability, as far as I can tell.
- Status
- Not open for further replies.