You're welcome.
By the way, function generators are great for frequency response testing because their level doesn't deviate as you change fine frequency (it's hard to make a Wien bridge flatter than 0.1dB), but their distortion is poor, typically 1%, and often visible on an oscilloscope as a pointy tip to sine waves, rather than a flat top. That's why I suggested a cheap and cheerful function generator - an expensive one will still distort.
By the way, function generators are great for frequency response testing because their level doesn't deviate as you change fine frequency (it's hard to make a Wien bridge flatter than 0.1dB), but their distortion is poor, typically 1%, and often visible on an oscilloscope as a pointy tip to sine waves, rather than a flat top. That's why I suggested a cheap and cheerful function generator - an expensive one will still distort.
If I could just interrupt... 
The driver stage is clipping, thus limiting the output stage. I played with some bias values for the Nuvistor, and even increased the voltage to 110V, and I still can only swing the same amount. I've reached the limit of the tube with that particular interstage. The Lundahl must be a bit more lossy than I thought, and have a bit less primary inductance, since I'm not getting a mu of 35 out of the Nuv. So... I settled on a cathode resistor of 150 ohms, everything else the same - and now I get 4 undistorted watts out of the amp.
Anyway, I thought about just scrapping the whole driver stage and going in a different direction - but I was enjoying the music too much.
It really sounds good. I may be becoming a push-pull guy?!?!? Regardless, I'll never go back to a SET that's down 3dB at 20kHz and 20Hz...
Bas - I added .047uF caps across the VR tubes - I can't see any difference on the scope, or hear any. Not even worth taking scope pictures of... I left them in there though - can't hurt.
Still haven't tracked down the hash source, but as I said, it's only 2mV.
Joel
The driver stage is clipping, thus limiting the output stage. I played with some bias values for the Nuvistor, and even increased the voltage to 110V, and I still can only swing the same amount. I've reached the limit of the tube with that particular interstage. The Lundahl must be a bit more lossy than I thought, and have a bit less primary inductance, since I'm not getting a mu of 35 out of the Nuv. So... I settled on a cathode resistor of 150 ohms, everything else the same - and now I get 4 undistorted watts out of the amp.
Anyway, I thought about just scrapping the whole driver stage and going in a different direction - but I was enjoying the music too much.
It really sounds good. I may be becoming a push-pull guy?!?!? Regardless, I'll never go back to a SET that's down 3dB at 20kHz and 20Hz...Bas - I added .047uF caps across the VR tubes - I can't see any difference on the scope, or hear any.
Not even worth taking scope pictures of... I left them in there though - can't hurt.Still haven't tracked down the hash source, but as I said, it's only 2mV.
Joel
James, you're right - an extra 3dB wouldn't make hardly any difference. 
And basically in order to get the full 15V swing on the output tubes, I'm realizing that I'd have to go with a Williamson topology - adding another gain stage before each output tube. Then I'd need some NFB... and more decoupling... and a bigger chassis... and...
No, I had no plans to plug guitars into this - mainly because I have at least 6 guitar amps clogging the apartment already.
cheers.
Joel
ps. The LL1660S is within 1mV between both 6AQ5 grids, at all the signal levels I measured it at. No AC balancing scheme is needed in this amplifier.
And basically in order to get the full 15V swing on the output tubes, I'm realizing that I'd have to go with a Williamson topology - adding another gain stage before each output tube. Then I'd need some NFB... and more decoupling... and a bigger chassis... and...
No, I had no plans to plug guitars into this - mainly because I have at least 6 guitar amps clogging the apartment already.
cheers.
Joel
ps. The LL1660S is within 1mV between both 6AQ5 grids, at all the signal levels I measured it at. No AC balancing scheme is needed in this amplifier.
I don't know why I didn't think of this before, but if I switch to 6BQ5's, and an 8k ohm p-p output transformer, I can easily drive the mere 8V needed for each grid. I'll get at the original spec'd 10 watts, without changing any voltages, or the driver. Distortion should actually go down too, since the EL84 has lower THD figures for the same output when compared directly to a 6AQ5/6V6.
I should be able to get them in by tuesday or so.
I should be able to get them in by tuesday or so.
I finished the change to EL84's, and Hammond 1608 output transformers. I also changed to fixed battery bias using a single 9V battery between the IT secondary center tap and ground.
I get 6 watts RMS.
Regulation however on the first 0A2/0B2 combination could be improved. I'm going to try a larger power tran. A 325V hammond model will probably be enough. The goal is to get more available voltage to drop across the first resistor - thus improving regulation. I tried cap input filter - too ugly. The chop definitely was making out to the speaker. So, I'd prefer to stay with a choke input.
To the nuvistor critics - I'm still eagerly waiting a viable alternative suggestion. It needs to swing 30v, with an Rp of 3k or less, an input of 1v or less, and an Ip of 10mA or less. Anyone?
I get 6 watts RMS.
Regulation however on the first 0A2/0B2 combination could be improved. I'm going to try a larger power tran. A 325V hammond model will probably be enough. The goal is to get more available voltage to drop across the first resistor - thus improving regulation. I tried cap input filter - too ugly. The chop definitely was making out to the speaker. So, I'd prefer to stay with a choke input.
To the nuvistor critics - I'm still eagerly waiting a viable alternative suggestion. It needs to swing 30v, with an Rp of 3k or less, an input of 1v or less, and an Ip of 10mA or less. Anyone?
Well, biasing it down to under half it's typical class A standing current is surely not going to help linearity.
It was a bit of a rhetorical question anyway. I consider the 6DV4 perfectly suited for this amp for me. If somebody wants to tackle this project and feels like they need the extra few watts they can squeeze out of the EL84 pair, then they should try other options - what those are, I don't know.
An easy answer is of course to switch to a modified Williamson. With the front end identical to the current configuration, one could add two additional triodes after the LL1660S, and drive either the 6BQ5's, or the more demanding 6AQ5's as in the original schematic. Then the Nuvistor would only be called on to swing a few volts at each driver's grid, which is quite easily done. I should think 2v into a 6J5 would produce very low distortion and enough gain. But of course it wouldn't be as low as the single Nuvistor. I really think these are ideal input tubes in many ways, when choke or IT loaded.
Joel
It was a bit of a rhetorical question anyway. I consider the 6DV4 perfectly suited for this amp for me. If somebody wants to tackle this project and feels like they need the extra few watts they can squeeze out of the EL84 pair, then they should try other options - what those are, I don't know.
An easy answer is of course to switch to a modified Williamson. With the front end identical to the current configuration, one could add two additional triodes after the LL1660S, and drive either the 6BQ5's, or the more demanding 6AQ5's as in the original schematic. Then the Nuvistor would only be called on to swing a few volts at each driver's grid, which is quite easily done. I should think 2v into a 6J5 would produce very low distortion and enough gain. But of course it wouldn't be as low as the single Nuvistor. I really think these are ideal input tubes in many ways, when choke or IT loaded.
Joel
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.