tubelab,
please ignore my previous post - my notes actually say 4w UL at a slightly lower op than yours. 5W sounds spot on.
And you are right, it does sound very nice.
pm
please ignore my previous post - my notes actually say 4w UL at a slightly lower op than yours. 5W sounds spot on.
And you are right, it does sound very nice.
pm
A recipe for success - I agree that should cope with transients most of the time, it's only if you want to amplify at sustained high volume that sag could be a problem. BTW, Johan, what rectifiers do you use?
I tried a pair of NOS 6550A's in a standard 50W fixed bias UL configuration with 5K a-a and got a lower pout compared to Svet 6550C's. Infact the latter behaved better with lower thd in AB2 i.e forced drive.
I went back to an old RCA data sheet, which specifies 3300 ohms for maximum transfer. Svet claim 4K a-a some even 5K. If clone manufacturers claim replica tube geometry, (which seems doubtful) then take your pick with actual load Z. This may explain some of the poor HF performance.
As with dummy loads, I use a standard non ind thick film resistor.
For electostatic LS sim, I use a series 20uH choke shunted by 2uFand 15 ohm res,
richj
I went back to an old RCA data sheet, which specifies 3300 ohms for maximum transfer. Svet claim 4K a-a some even 5K. If clone manufacturers claim replica tube geometry, (which seems doubtful) then take your pick with actual load Z. This may explain some of the poor HF performance.
As with dummy loads, I use a standard non ind thick film resistor.
For electostatic LS sim, I use a series 20uH choke shunted by 2uFand 15 ohm res,
richj
Rectifiers are BY228 - 1.5KV 3A in bridge. My input is 660VAC. (Not a lot of options here for rectifiers of that rating.) That does however give me 105W output with rms input (rest is stabilised at 500V, so not that much effect from sag).
I have concocted an l.s. equivalent load as said but just to check (it is not 100W!). Load for max. output is an 8 ohm 20W resistor in a bucket of water!
(Now why do I think the coffee made from audio heated water tastes better ...... oops, sorry.)
I have concocted an l.s. equivalent load as said but just to check (it is not 100W!). Load for max. output is an 8 ohm 20W resistor in a bucket of water!
(Now why do I think the coffee made from audio heated water tastes better ...... oops, sorry.)
I have been using a pair of Radio Shack 20W 8 ohm "non - ind" resistors in a coffee cup full of water for about 3 years at power levels up to 100 watts. One of them is all cracked and discolored. It still works fine. The other one ( no cracks or discoloration) was loading a 15 watt SE amp (without water) and it just went open with no warning. Nothing bad happened this time. I think that this would have been different if I was testing a 100 watt P-P amp.
I bought a bunch of 1 ohm 30 watt thin film resistors. I will make two strings of 10 mounted on a large heat sink. It is cheaper than a fried OPT and can give me loads in 1 ohm steps.
Ray and Johan, you are right.... a good screen supply is not expensive or difficult. I have had decent results with R/C on the screen with good capacitance. Kinda makes me wonder why I bother with UL now....
AAArghhh!!!
Now what do I say to that, after all this
🙂
And then the man also likes toy trains to boot! (Mine is HO)
Now what do I say to that, after all this
🙂 And then the man also likes toy trains to boot! (Mine is HO)
With power pentodes/tetrodes, a stable and low-impedance power source is needed for the screens. What size cap and resistor do you mean?
You too, huh?
I think I used a 10 uF on the screen, I don't know what would be optimum, but that should hold it pretty steady. Hmm, maybe I should have bypassed it with a film cap too.
Why not UL??
How about better top end performance with less expensive transformers. And, since it is just one form of negative feedback, could it be better to delete it and do this: Optimise a straight pentode stage, then add a little local negative feedback from the transformer output (instead of primary in UL) to grid 1 instead of screen grid. That way, you correct transformer too.
What do think?
John, I was into HO a long time ago, but after having kids, I found Lionel to be better match for our purposes. Even though the kids are grown now, I have been having fun with it ever since!
Why not UL??
How about better top end performance with less expensive transformers. And, since it is just one form of negative feedback, could it be better to delete it and do this: Optimise a straight pentode stage, then add a little local negative feedback from the transformer output (instead of primary in UL) to grid 1 instead of screen grid. That way, you correct transformer too.
What do think?
John, I was into HO a long time ago, but after having kids, I found Lionel to be better match for our purposes. Even though the kids are grown now, I have been having fun with it ever since!
Kent Smith said:
You need a driver with less output resistance. Much less than for a power transfer because impedance of the output stage will be non-linear.
Hi Kevin,
"Why not pentodes?"
1. A little difficult to follow why you say better top end performance. In my experince I never had a problem with that in UL.
2. Despite whatever NFB round the "outside' of a pentode stage, the tube itself still sees a varying output load; it will still present pentode varying distortion into a non-constant load. NFB will drop the lot, but the nature will still be varying with high order distortion. Triode-equivalent performance obviates that.
(In this respect there was a valuable article by Prof Otala et al in Wireless World Nov. & Dec. 1980, on Intermodulation at the amplifier-loudspeaker interface. Though addressing ss topology, the principles are universal. Briefly, it showed that as a result of high-Z in the output devices themselves, reactive energy stored in the loudspeaker can reflect back there and further, with detrimental overall effect. This is not corrected by NFB as would be the case with a low-Z output device, e.g. in our case, a triode.)
3. The UL output configuration is an intrinsic output thing, not the same as when involving G1 as well. It has inherent low output impedance, as stated here before. Distortion graphs also show this as showed before in this thread.
4. More expensive output transformers? The Hammond site was not kind to me just now and I could not find costs; you will have to inform me of that aspect. As said, having them wound locally involves an insignificant increase in cost.
This apart from the cathode winding variation; that adds extra benefits at the expense of higher g1 signal amplitude.
I would suggest that your G2 bypass capacitor of 10 uF might be low. Keep in mind that g2 is not a (high-Z) control grid; since it draws current it is rather another anode, with an equivalent "rp". One is looking at a finite and greatly varying "r.g2"; thus the bypass may need to be more. Best is to look there with a scope while the amp is playing music (slow time scale).
Regards
______________________________________________-
Ah! Lionel!
Yet again, economy rules my desires.
"Why not pentodes?"
1. A little difficult to follow why you say better top end performance. In my experince I never had a problem with that in UL.
2. Despite whatever NFB round the "outside' of a pentode stage, the tube itself still sees a varying output load; it will still present pentode varying distortion into a non-constant load. NFB will drop the lot, but the nature will still be varying with high order distortion. Triode-equivalent performance obviates that.
(In this respect there was a valuable article by Prof Otala et al in Wireless World Nov. & Dec. 1980, on Intermodulation at the amplifier-loudspeaker interface. Though addressing ss topology, the principles are universal. Briefly, it showed that as a result of high-Z in the output devices themselves, reactive energy stored in the loudspeaker can reflect back there and further, with detrimental overall effect. This is not corrected by NFB as would be the case with a low-Z output device, e.g. in our case, a triode.)
3. The UL output configuration is an intrinsic output thing, not the same as when involving G1 as well. It has inherent low output impedance, as stated here before. Distortion graphs also show this as showed before in this thread.
4. More expensive output transformers? The Hammond site was not kind to me just now and I could not find costs; you will have to inform me of that aspect. As said, having them wound locally involves an insignificant increase in cost.
This apart from the cathode winding variation; that adds extra benefits at the expense of higher g1 signal amplitude.
I would suggest that your G2 bypass capacitor of 10 uF might be low. Keep in mind that g2 is not a (high-Z) control grid; since it draws current it is rather another anode, with an equivalent "rp". One is looking at a finite and greatly varying "r.g2"; thus the bypass may need to be more. Best is to look there with a scope while the amp is playing music (slow time scale).
Regards
______________________________________________-
Ah! Lionel!
Yet again, economy rules my desires.
I believe the load sensitivity of pentodes can indeed be significantly reduced by the application of local NFB. See Designing Your Own Amplifier Pt. 5: Feedback Amplifiers by Norman Crowhurst. His use of balanced local NFB loops from OP tube plates to driver cathodes was aimed at reducing load variation sensitivity significantly.
Again, a lower well regulated screen grid supply AND a separate UL windings is an interesting option to try! 
For example, using an another smaller output transformer with secondaries connected to speaker taps of the main output transformer, just for experimenting...
For example, using an another smaller output transformer with secondaries connected to speaker taps of the main output transformer, just for experimenting...
Wavebourn said:
If going to all the trouble of a second transformer, put it in the cathode circuit!
rdf said:
If going to all the trouble of a second transformer, put it in the cathode circuit!
No harm in experimenting (C) Wavebourn 😉
Wavebourn said:
Anatoliy!!!!!
What are you trying to do? All those losses and leakage reactances! Are you that heavily against simple old UL?
ray_moth said:
Ah yes! The worthy Mr Crowhurst!
I read the lot again just as a treat to read proper design procedures. But I did say that such NFB would reduce pentode distortion; I was wondering about the nature of it remaining, i.e. quite variable distortion, etc., if you look back at pt. 2 of my previous post.
This is however getting less important - if all is reduced, is that not enough? I was simply thinking of loudspeaker reactive feedback still getting in the way as per the Otala article. But over one stage only - I am not going to seriously argue there. As an alternative to UL (per topic) ..... I still cannot see my using pentode-with-feedback. And by the time we have a second G2 feedback winding, one might as well go Quad II. (Of course, such transformers are not commercially available to my knowledge.)
I am still trying to find why not UL; the only reason I heard was about the more expensive transformer (was it Kevin?) - if that much more expensive.
But thanks for "bringing back" N.C.C.
Don't take good ole Norm as some sort of fount from which all wisdom spews. His description of PP load lines is about as cloudy as they come...not to mention a bit contradictory at times.
cheers,
Douglas
cheers,
Douglas
- Status
- Not open for further replies.