Finally the 10Y. Now this is really hard to criticise. Good right down from upper treble to bass. Smooth, very detailed. What is not to like. It's a better match for the 126C so I think this combination is a result. The 49 was just a passing fancy and ultimately the treble was a fraction papery on violins though very clear - maybe a different circuit?
Does the 10Y displace the 26? That's a tough one. I wouldn't like to say. You could be happy with either, depending on the implementation. And also probably the 01A which has its own unique charms.
Andy
Does the 10Y displace the 26? That's a tough one. I wouldn't like to say. You could be happy with either, depending on the implementation. And also probably the 01A which has its own unique charms.
Andy
The 126C is probably borderline for a 26 - more than 106 Henries would be good. But It's what I'm listening to and I'm more than OK with the sound. At least it means you don't have to spring for Lundahl LL1660 which is expensive.
Next experiment is filament bias. I'll report back as soon as I try it. It needs a bit of calculation, but I've put it on my Excel spreadsheet, which is what I do with all by tube calculations (of which there are many!!!).
Filament bias is fine for a 26. More problematic for a 10Y though can be done. For a 26, if you use a 12+12v secondary and a LM1085 voltage reg, you need below about 12 volts for the bias so as to stay in regulation, which suits a 26 at 6ma just fine. It would also suit a 01A.
Andy
Next experiment is filament bias. I'll report back as soon as I try it. It needs a bit of calculation, but I've put it on my Excel spreadsheet, which is what I do with all by tube calculations (of which there are many!!!).
Filament bias is fine for a 26. More problematic for a 10Y though can be done. For a 26, if you use a 12+12v secondary and a LM1085 voltage reg, you need below about 12 volts for the bias so as to stay in regulation, which suits a 26 at 6ma just fine. It would also suit a 01A.
Andy
OK - filament bias. I used a 9 ohm cathode resistor made up of 3 big WW 27 ohms in parallel (this gets hot with 1 amp through it). I used 150v on the plate from a Glow tube. Supply voltage to the filament bias was 10.1v, which comes from a bench voltage supply (Thurlby Thandar).
This is very good - clean clear sound and vocals are magical. I could probably do with more inductance than the 126C can provide - something Lundahl no doubt.
I think this is the best yet. It's a more forward and involving sound than the 10Y (?would be less forward with more inductance?). But I haven't tried the 10Y in filament bias yet.
Andy
This is very good - clean clear sound and vocals are magical. I could probably do with more inductance than the 126C can provide - something Lundahl no doubt.
I think this is the best yet. It's a more forward and involving sound than the 10Y (?would be less forward with more inductance?). But I haven't tried the 10Y in filament bias yet.
Andy
Yes - the last post was filament bias with the 26.
I'm just listening to filament bias with the 10Y. In this setup I have a 5 ohm cathode resistor (which needs to be rated 25 watts) and again 150v on the plate though a glow tube. Supply voltage in this case is 13v. Supply voltage varies with the tube - one's on 12.8 and the other on 13.5 for 1.2A current.
It's really hard to choose between the 26 and 10Y. I find the 10y possibly a bit smoother and the 26 more involving. No doubt this is also the interaction with the 126C transformer and the operating points. But they're both good.
Andy
I'm just listening to filament bias with the 10Y. In this setup I have a 5 ohm cathode resistor (which needs to be rated 25 watts) and again 150v on the plate though a glow tube. Supply voltage in this case is 13v. Supply voltage varies with the tube - one's on 12.8 and the other on 13.5 for 1.2A current.
It's really hard to choose between the 26 and 10Y. I find the 10y possibly a bit smoother and the 26 more involving. No doubt this is also the interaction with the 126C transformer and the operating points. But they're both good.
Andy
Lastly filament bias with the 01A. I used 150v glow tube for the plate voltage (bit high, but it's temporary). 50 ohms cathode resistor should give 10v bias for 3.5mA. This should be run less hot but that's the setup I have easily available.
Sound is great. Right up there with 26 and 10Y. No doubt about that. I think of the 3 I'd give it to the 26 by a whisker. But they're all really good.
Andy
Sound is great. Right up there with 26 and 10Y. No doubt about that. I think of the 3 I'd give it to the 26 by a whisker. But they're all really good.
Andy
hello Andy, I tried the filament bias trick when experimenting with DHT drivers for my 300B-SE. I found that it sounded best in this setup when the cathode resistor was given 300mA from the same supply as the filament. (Rather than the full 1A)
It's easy to do using a PNP transistor source.
I've been redesigning that old DHT heater circuit to give lower noise, higher ripple rejection, and automatic programming of the current value to achieve the correct voltage across the filament. Maybe I should add the extra current source in there when I translate it to PCB. It could be offered as a kit including the hard-to-get transistors that make it work best.
It's easy to do using a PNP transistor source.
I've been redesigning that old DHT heater circuit to give lower noise, higher ripple rejection, and automatic programming of the current value to achieve the correct voltage across the filament. Maybe I should add the extra current source in there when I translate it to PCB. It could be offered as a kit including the hard-to-get transistors that make it work best.
Hi Rod,
I don't know enough about solid state to understand the circuit you're suggesting to get the 300mA, but interested!
I'll be using filament bias in future - it just sounds so incredibly clean. So this does call into question how I provide the filament DC. Thomas Mayer said he didn't like current sources for this and preferred a choke. One thing that's needed is some way of adjusting the bias quite precisely. right now I'm using bench supplies with a current indicator, but when I build a supply I'll be wanting to find how best to get the voltage/current right. I think my first sttempt will be just a LM1085 as voltage reg then see where to go from there.
I'm wondering if it would be convenient to put some capacitors and Shottky diodes on the PCB - make it self contained.
Andy
I don't know enough about solid state to understand the circuit you're suggesting to get the 300mA, but interested!
I'll be using filament bias in future - it just sounds so incredibly clean. So this does call into question how I provide the filament DC. Thomas Mayer said he didn't like current sources for this and preferred a choke. One thing that's needed is some way of adjusting the bias quite precisely. right now I'm using bench supplies with a current indicator, but when I build a supply I'll be wanting to find how best to get the voltage/current right. I think my first sttempt will be just a LM1085 as voltage reg then see where to go from there.
I'm wondering if it would be convenient to put some capacitors and Shottky diodes on the PCB - make it self contained.
Andy
When you compare choke-fed CCS to solid state, please remember to take account of noise. The LT1086 is a better 3-pin regulator than most, but its data sheet still specifies
'Noise rms, 10Hz..10kHz 0.003% of Vout.' Sounds small, but at 5V that's 150uV, and it'll be worse than that because you're certainly not band-limited to 10kHz. And that's rms as well, so the peaks will be much larger.
If the output of this chip is injected across the cathode resistor, you're effectively putting the whole noise signal at the cathode-to-ground of the valve, which is the same as injecting it to the grid!
150uV is only 35dB or so down on line-level signals, so you'll hear the noise all the time, and it will be sure to degrade the sound.
When comparing solid-state CCS to choke-fed, a low-noise CCS will need to be built - but that's not very tough. The 2-transistor CCS shown in my old heater circuit can be adapted: just use a low-noise transistor for the feedback position (BC547C, etc), and use a low value of collector resistor in it, so that the BC547C Ic is at least 10mA. Most ZTX transistors are very low-noise if run at 10..40mA too (low enough for a moving-coil cartridge). I'll draw it when I get a minute.
Point noted about adding rectifiers! thanks Andy.
'Noise rms, 10Hz..10kHz 0.003% of Vout.' Sounds small, but at 5V that's 150uV, and it'll be worse than that because you're certainly not band-limited to 10kHz. And that's rms as well, so the peaks will be much larger.
If the output of this chip is injected across the cathode resistor, you're effectively putting the whole noise signal at the cathode-to-ground of the valve, which is the same as injecting it to the grid!
150uV is only 35dB or so down on line-level signals, so you'll hear the noise all the time, and it will be sure to degrade the sound.
When comparing solid-state CCS to choke-fed, a low-noise CCS will need to be built - but that's not very tough. The 2-transistor CCS shown in my old heater circuit can be adapted: just use a low-noise transistor for the feedback position (BC547C, etc), and use a low value of collector resistor in it, so that the BC547C Ic is at least 10mA. Most ZTX transistors are very low-noise if run at 10..40mA too (low enough for a moving-coil cartridge). I'll draw it when I get a minute.
Point noted about adding rectifiers! thanks Andy.
Member
Joined 2006
With filament bias and a 10 ohm cathode resistor with 10 watts through it, I need a big part like 30W or over - they get hot.
What would you guys suggest for the best sounding part here? I tried alu clad, which are very clear but rather hard, cheap ceramic WW which are nothing special but less hard at least.
what's the best sound here? I see some Vishay Sfernice 30W thick film parts for instance. Another option is those brown ceramic Ohmite WW resistors. I don't know if those dark green ceramic WW types (?make) go down as far as 10 ohms.
Ideas?
Andy
What would you guys suggest for the best sounding part here? I tried alu clad, which are very clear but rather hard, cheap ceramic WW which are nothing special but less hard at least.
what's the best sound here? I see some Vishay Sfernice 30W thick film parts for instance. Another option is those brown ceramic Ohmite WW resistors. I don't know if those dark green ceramic WW types (?make) go down as far as 10 ohms.
Ideas?
Andy
The dark green WW are Welwyn Vitreous enamelled WW.
You can go 10R/14W -
http://uk.farnell.com/welwyn/w24-10r-ji/resistor-ww-14w-5-10r/dp/9505580
Edit: watch out, these are inductive, and have magnetic endcaps! Try them though, they are low noise, and if they are fed from low noise source, and kept away from stray fields (trafos, rectifier wiring etc) they work well.
You can go 10R/14W -
http://uk.farnell.com/welwyn/w24-10r-ji/resistor-ww-14w-5-10r/dp/9505580
Edit: watch out, these are inductive, and have magnetic endcaps! Try them though, they are low noise, and if they are fed from low noise source, and kept away from stray fields (trafos, rectifier wiring etc) they work well.
Last edited:
OK - here's some cathode resistor shootouts. Part is a 10 ohm resistor which needs 10 watts. So this must be over-rated for comfort.
Multicomp vitreous enamel 10R 50W
This is a big resistor, 100mm long. Heat dissipation is fine. This was the best sound - very clean with lots of life and subtle tone qualities to instruments. Vocals were lively and general sound smooth but vivid.
NEOHM Cement WW - 3 of 27R in parallel rated 10w each, total 30W
Heat dissipation is also OK here. Sound is neutral, good bass and treble. Easy to listen to. Not as lively or detailed as the above, but nothing objectionable.
Alu Clad Vishay 10R 50W
Very clear and detailed but a rather hard sound which gets tiring. This may be better with more damping - I had it in free air. Not an option unless it can be made smoother.
Vishay thick film TO-220 case 10R 50W on heatsink 3.9 C/W
I expected a lot from this, and was disappointed. Sound was dead and lifeless, lacked bass extension, and vocals were shut in. It was smoother than the alu clad but overall inferior to even the cement WW. Could do with a bigger heatsink, though 3.9 C/W was OK for testing.
So the conclusion is clearly to use vitreous enamel wirewounds in some form. For a 10 ohm cathode resistor which suite the 26 in filament bias, 50W is a good rating, and 30W possible for mounting on top of the case. I've had numerous suggestions to use Mills 12W resistors in parallel. Three of the 30, four of the 39 or five of the 47. Not cheap!! Also not available at good prices in the UK - they are £4 here.
Andy
Multicomp vitreous enamel 10R 50W
This is a big resistor, 100mm long. Heat dissipation is fine. This was the best sound - very clean with lots of life and subtle tone qualities to instruments. Vocals were lively and general sound smooth but vivid.
NEOHM Cement WW - 3 of 27R in parallel rated 10w each, total 30W
Heat dissipation is also OK here. Sound is neutral, good bass and treble. Easy to listen to. Not as lively or detailed as the above, but nothing objectionable.
Alu Clad Vishay 10R 50W
Very clear and detailed but a rather hard sound which gets tiring. This may be better with more damping - I had it in free air. Not an option unless it can be made smoother.
Vishay thick film TO-220 case 10R 50W on heatsink 3.9 C/W
I expected a lot from this, and was disappointed. Sound was dead and lifeless, lacked bass extension, and vocals were shut in. It was smoother than the alu clad but overall inferior to even the cement WW. Could do with a bigger heatsink, though 3.9 C/W was OK for testing.
So the conclusion is clearly to use vitreous enamel wirewounds in some form. For a 10 ohm cathode resistor which suite the 26 in filament bias, 50W is a good rating, and 30W possible for mounting on top of the case. I've had numerous suggestions to use Mills 12W resistors in parallel. Three of the 30, four of the 39 or five of the 47. Not cheap!! Also not available at good prices in the UK - they are £4 here.
Andy
Andy, I assume the Vitreous part was
MULTICOMP|MC14663|RESISTOR, 10 OHM 50W | Farnell United Kingdom
Thick Film parts do show higher current noise, and since you're running high current, they may fail due to noise. Don't let it put you off trying them in low-current high voltage jobs!
The Almi-clad types usually have steel terminals, which won't help there. WW does do best for current noise, so perhaps a noninductive (wound in a Ayrton-Perry stylee) will be best. I believe Ohmite make a good choice of these.
MULTICOMP|MC14663|RESISTOR, 10 OHM 50W | Farnell United Kingdom
Thick Film parts do show higher current noise, and since you're running high current, they may fail due to noise. Don't let it put you off trying them in low-current high voltage jobs!
The Almi-clad types usually have steel terminals, which won't help there. WW does do best for current noise, so perhaps a noninductive (wound in a Ayrton-Perry stylee) will be best. I believe Ohmite make a good choice of these.
The Vitreous resistor was indeed Farnell's Multicomp part. They had five, and I took four. Next delivery is in September, so I'll be looking around. A vitreous enamel WW rated 50w is the part to go for, so it seems at present. Will be seeking out some Mills 12W as and when I see them.
With such a simple preamp, every part is audible. I was surprised at the difference these resistors make.
Andy
With such a simple preamp, every part is audible. I was surprised at the difference these resistors make.
Andy