The dynamic output impedance of your supply is higher than with a VR tube alone, and will depend entirely on the 3K3 resistor and 10uF capacitor. Plot that vs frequency and see what happens. If you want more filtering do it before the VR tube. But then the noise of the VR tube is something you'll have to accept and live with.
IKO:
The VR with a CCS performs really well, noise is below 85dB. This is why Gary Pimm started using them in his circuits:
http://www.bartola.co.uk/valves/2012/08/16/russian-voltage-regulator-test/
Ale
The VR with a CCS performs really well, noise is below 85dB. This is why Gary Pimm started using them in his circuits:
http://www.bartola.co.uk/valves/2012/08/16/russian-voltage-regulator-test/
Ale
Ale
Agreed Gary often uses a shunt reg/CCs combination but would never place an RC combination between them as this would negate the tight regulation achieved by the combination. See his schematic for the 85 linestage.
the exception was when he deliberately degraded regulation to the screens of the 6AU6s in theTabor using a series resistor after the shunt regs because he felt it sounded better that way. But this was due to a complex interaction with the rest of the circuit.
As iko stated, If you need more filtration place it before the vr tubes.
However, it may well be that the advantage you get from using the rc combo to get the exact operating point you want outweighs the disadvantage you get from loss of regulation. Your call.
Agreed Gary often uses a shunt reg/CCs combination but would never place an RC combination between them as this would negate the tight regulation achieved by the combination. See his schematic for the 85 linestage.
the exception was when he deliberately degraded regulation to the screens of the 6AU6s in theTabor using a series resistor after the shunt regs because he felt it sounded better that way. But this was due to a complex interaction with the rest of the circuit.
As iko stated, If you need more filtration place it before the vr tubes.
However, it may well be that the advantage you get from using the rc combo to get the exact operating point you want outweighs the disadvantage you get from loss of regulation. Your call.
VR tube noise is a non-issue in my 26 preamp, I can't hear anything at the speaker. Having a cascoded CCS before and after the VR tubes and a small bypass cap across the VR tubes seems to work just fine, and I'm even using the mu output of the second CCS, which supposedly degrades the PSRR. Combine that with the warm glow and it's a winner in my book!
Last edited:
Ale, I just stated the fact. Whether a higher output impedance is of any consequence in your preamp that's another matter. Most likely not noticeable.
Magz, in my older implementation with the 156C plate choke I could clearly hear the VR noise as a hiss when no music was playing and the volume was way up. Didn't bother me much; and the warm glow was a winner in my book too.
Magz, in my older implementation with the 156C plate choke I could clearly hear the VR noise as a hiss when no music was playing and the volume was way up. Didn't bother me much; and the warm glow was a winner in my book too.
hi guys,
thanks for suggestions.
Will try to find those rubber suspensions at a local hardware shop or will study a diy solution like these shared.
Anto
Ale, I just stated the fact. Whether a higher output impedance is of any consequence in your preamp that's another matter. Most likely not noticeable.
Magz, in my older implementation with the 156C plate choke I could clearly hear the VR noise as a hiss when no music was playing and the volume was way up. Didn't bother me much; and the warm glow was a winner in my book too.
In my first design I used a VR150/0D3 with plate to line transformers and in this implementation the noise was plainly audible, while it looked cool it seriously compromised the overall system SNR.. Obviously this is not going to be a big issue with para-feed, choke or CCS loaded outputs. (I'm assuming perhaps wrongly that a good audio choke having relatively low inter-winding capacitance should be adequate, or you could put a couple in series.)
On the subject of microphonics I have been using a pair of Sylvania ST type 26s that while microphonic seemed to be better than a lot of others I have in my stash. I think these are starting to go after about six years of relatively hard use, and installed a NOS well matched pair of RCAs which proved to be incredibly microphonic and one of which also generated extraneous noise.. (Unfortunately I have a fair number of NOS RCAs.. )
I've found that wrapping the envelope of ST types starting about 1cm up from the base and all the way up to the mica with wide (1.5cm) plumbers teflon tape makes a surprising improvement in microphony in many cases. I generally apply two layers heavily overlapped (50% or more) starting at the bottom and continuing to the top and then back down again in a single wrap.
A quick search on eBay last night indicated that there are comparatively few NOS 26 at the moment. The supply definitely seems much less plentiful than it did more than a decade ago when I designed my first one. Has me thinking about finding an alternative.
I've found that wrapping the envelope of ST types starting about 1cm up from the base and all the way up to the mica with wide (1.5cm) plumbers teflon tape makes a surprising improvement in microphony in many cases. I generally apply two layers heavily overlapped (50% or more) starting at the bottom and continuing to the top and then back down again in a single wrap.
A quick search on eBay last night indicated that there are comparatively few NOS 26 at the moment. The supply definitely seems much less plentiful than it did more than a decade ago when I designed my first one. Has me thinking about finding an alternative.
Vibration
Try E.A.R. grommets: E-A-R -- noise control, vibration damping, sound insulation, shock cushioning
McMaster-Carr sell this type: McMaster-Carr
.... but not outside of States.
"Due to the complexity of U.S. export regulations, McMaster-Carr accepts international orders only from our established customers. This decision also applies to orders shipping within the United States, because it is based on the final destination of the items. We cannot accept this order or future orders." :-(
Try E.A.R. grommets: E-A-R -- noise control, vibration damping, sound insulation, shock cushioning
McMaster-Carr sell this type: McMaster-Carr
.... but not outside of States.
"Due to the complexity of U.S. export regulations, McMaster-Carr accepts international orders only from our established customers. This decision also applies to orders shipping within the United States, because it is based on the final destination of the items. We cannot accept this order or future orders." :-(
Ale, I just stated the fact. Whether a higher output impedance is of any consequence in your preamp that's another matter. Most likely not noticeable.
Magz, in my older implementation with the 156C plate choke I could clearly hear the VR noise as a hiss when no music was playing and the volume was way up. Didn't bother me much; and the warm glow was a winner in my book too.
Did you feed glow tube with a CCS? That is the only way you will get no noise (85dB rejection over the raw supply) from the glow stage.
Regarding the output impedance. Looking from the transformer side, the 10u cap is in parallel with 3k3 in series with the glow tube. The cap impedance is about 800 ohms at 20Hz. If neglecting the VR impedance then the equivalent is somewhere below 800 ohms which I can't see that negative for a preamp with quiescent current of less than 6mA...
Ale
similar Anti-vibration bushings are available in Europe too (Germany) ...
https://www.buerklin.com/default.asp?event=ShowArtikel(12H3788)&l=e&jump=ArtNr_12H3788
there you can also find threaded metal blocks, similiar to those Thomas Mayers is using for his excellent idea/design for suspending sub chassis ...
https://www.buerklin.com/default.asp?event=ShowArtikel(17H912)&l=e&jump=ArtNr_17H912
I'm talking about the intrinsic noise of the VR tube, which is not dependent on what the VR tube is fed by.
As long as you're happy with that, I'm happy.
Only reason why I added the RC is to provide the necessary voltage drop to achieve the operating point I wanted. I will measure noise level when I get the chance to do so and will also test removing the RC. Bringing up the operating point for 30-40V won't make a big difference though.
I think if I don't experiment here I won't learn or potentially discover other ways of improving things. So far I have been doing everything that everyone else recommended here (and in other posts): filament bias, gyrators, OT, DC filament regulation, etc.
Cheers, Ale
I think if I don't experiment here I won't learn or potentially discover other ways of improving things. So far I have been doing everything that everyone else recommended here (and in other posts): filament bias, gyrators, OT, DC filament regulation, etc.
Cheers, Ale
Ale, in my opinion the entire community has benefited from your experiments. Great exploration!
Why not use a VR tube and a zener below it? 1N4757A (51V) or 1N4758A (56V) so you can get 126V or 131V out respectively? The zener can be bypassed with a larger cap if you want to muffle its noise.
Why not use a VR tube and a zener below it? 1N4757A (51V) or 1N4758A (56V) so you can get 126V or 131V out respectively? The zener can be bypassed with a larger cap if you want to muffle its noise.
I'm with the disco-verer: To notice or learn, especially by making an effort.
What cable arrangement are people using for outboard DC filament supplies? I don't know if this topic has been covered - it's such a long thread. For connectors I use 4 pin XLRs and currently just twisted unscreened pairs of 1 metre length. I have 10,000uF capacitors inside the main signal chassis
So what kind of cable is best for this job? The DC supply in filament bias is actually in the signal path at the tube socket, though the supply is floating. So low noise and hum should be a priority. So what do I consider here:
- twisted pairs
- twisted pairs with screen
- 2 twisted pairs
- 2 twisted pairs in one screen
And so on....
Also, should the screen be connected at one end (which?) or both?
All recommendations appreciated!
Andy
So what kind of cable is best for this job? The DC supply in filament bias is actually in the signal path at the tube socket, though the supply is floating. So low noise and hum should be a priority. So what do I consider here:
- twisted pairs
- twisted pairs with screen
- 2 twisted pairs
- 2 twisted pairs in one screen
And so on....
Also, should the screen be connected at one end (which?) or both?
All recommendations appreciated!
Andy
Hi Andy, the actual implementation will depend on the components you use, and their location. My thoughts:
A 2-chassis solution is easier to make quiet. If you have 2 chassis, put all of the mains transformers on the power chassis (none at all on the valve chassis). All mains transformers will leak a little current into their frames, and any stray field from them will induce small eddy-current in the chassis. The effect may not be big, but if you want the best, it may add up to something!
The power supply chassis should also have the rectifier, and big reservoir caps. If these are mounted on the signal chassis, the big recharge current-pulses will transmit EM field all around the room!
The Regulator (I assume you always use them!) is best mounted on the signal chassis. This way, the filament exposure to common-mode pickup is minimized.
It only remains to connect the raw dc supply to the regulators. Twisting the cable will reduce differential-mode pickup, but with my regulators, differential mode noise will be highly rejected anyway.
Shielding the cables helps reject common-mode pickup, and so is desirable. common-mode rejection can also be improved by mounting a common-mode choke on the signal chassis (away from any choke or OPT), and shunting the noise to the chassis (using e.g 47nF 630V MKP stacked-type, or for +, one for -). In this case, connect the cable shield at both ends, and maybe even use a clamp-on EMC ferrite on the cable. Or, you could connect the 2 chassis together using some wide braided copper.
The safety earth should be attached to the power chassis by means of a bolt/nut that is not used to mount any other component or bracket etc.
A separate conductor should be similarly mounted on the signal chassis and connected to the safety earth. An EMC ferrite on this last cable may also be worth trying.
The HT B+ supply must be in a separate cable, preferably shielded, using cable and connectors rated for 440V ac rms.
Use fuses for both supplies, mounted in the PSU chassis.
A 2-chassis solution is easier to make quiet. If you have 2 chassis, put all of the mains transformers on the power chassis (none at all on the valve chassis). All mains transformers will leak a little current into their frames, and any stray field from them will induce small eddy-current in the chassis. The effect may not be big, but if you want the best, it may add up to something!
The power supply chassis should also have the rectifier, and big reservoir caps. If these are mounted on the signal chassis, the big recharge current-pulses will transmit EM field all around the room!
The Regulator (I assume you always use them!) is best mounted on the signal chassis. This way, the filament exposure to common-mode pickup is minimized.
It only remains to connect the raw dc supply to the regulators. Twisting the cable will reduce differential-mode pickup, but with my regulators, differential mode noise will be highly rejected anyway.
Shielding the cables helps reject common-mode pickup, and so is desirable. common-mode rejection can also be improved by mounting a common-mode choke on the signal chassis (away from any choke or OPT), and shunting the noise to the chassis (using e.g 47nF 630V MKP stacked-type, or for +, one for -). In this case, connect the cable shield at both ends, and maybe even use a clamp-on EMC ferrite on the cable. Or, you could connect the 2 chassis together using some wide braided copper.
The safety earth should be attached to the power chassis by means of a bolt/nut that is not used to mount any other component or bracket etc.
A separate conductor should be similarly mounted on the signal chassis and connected to the safety earth. An EMC ferrite on this last cable may also be worth trying.
The HT B+ supply must be in a separate cable, preferably shielded, using cable and connectors rated for 440V ac rms.
Use fuses for both supplies, mounted in the PSU chassis.
Hi Rod!
There's a lot to digest there. Right now I just want to wire up a few cables for my filament supplies. So looking at cable types. My filament supplies may go up to about 2 amps but usually not over 1.5A so first decision is what thickness wire. I'm thinking 20AWG is sufficient - not sure about 22AWG.
Then if I ground the shield one end this is looking more like a 5 pin XLR. Easier for me to put this at the PSU end since all my amp chassis are wired up with 4 pin XLRs. Just ground the screen one end.
How necessary is the screen? And if it is necessary, then is grounding it at the PSU adequate?
Just trying to get the simple decisions right first. Just filaments here, not HT.
Andy
There's a lot to digest there. Right now I just want to wire up a few cables for my filament supplies. So looking at cable types. My filament supplies may go up to about 2 amps but usually not over 1.5A so first decision is what thickness wire. I'm thinking 20AWG is sufficient - not sure about 22AWG.
Then if I ground the shield one end this is looking more like a 5 pin XLR. Easier for me to put this at the PSU end since all my amp chassis are wired up with 4 pin XLRs. Just ground the screen one end.
How necessary is the screen? And if it is necessary, then is grounding it at the PSU adequate?
Just trying to get the simple decisions right first. Just filaments here, not HT.
Andy