Last week I did some preliminar tests with the LL2746 in 1:2 step-up mode. Despite having measured good results with it, it will be a challenge to drive grid current given that the output impedance of the 4P1L will be multiplied by 4 so about 5KΩ.
Before looking at the LL1671/20mA which is suitable for multiple driver valves, let’s see how the LL2746 driver performs with the addition of the input step-up microphone transformer LL7903. I’m currently using the LL7903 in my 814 SE A2 amplifier and sounds really nice. The LL7903 was wired up in 1:4 setup so gain can get about 63:
The addition of the LL7903 doesn’t impact negatively the performance of the driver. In fact, at full swing of 200Vpp the THD is consistent at about 0.5%. The secondary of the LL7903 doesn’t include any zobel compensation so we can see the 30kHz peak in the frequency response.
Now, let’s look at the LL1671. I have one gapped at 20mA, so the driver operating point has to change in order to allow full 200Vpp swing without clipping. I changed the filament resistor to 25Ω and the set the variable power supply to the following operating point:
The performance is better than the previous circuit:
Compared to the LL2746, the gain now has a reduced gain of about 36. Still the gain is as good as an average triode-strapped pentode/tetrode. The overall THD has been reduced to about 0.3% instead of the 0.5% we had before. The frequency response shows a clear peak at about 50kHz now as neither the LL7903 or the LL1671 has a Zobel compensation. The -3dB corner is roughly the same at about 10Hz. The LL1671 gapped at 20mA has a slightly higher primary inductance (5H difference estimated), however the 10Ω increase in the filament resistor increases the anode dynamic impedance by 100Ω approximately. In summary the net effect of change in L and R makes the pole to stay in the same place 🙂
This final circuit seems like a reasonable compromise to use the 4P1L DHT driver in filament bias to deliver 200Vpp drive at low distortion and low impedance. An LL1692a in (2:3.5) could be investigated, but I don’t have the transformer to do the tests 🙁
What I didn’t do was to compare the LL1671 in Alt T mode (1:2) which should be similar to the LL2746.
Excellent work Ale. One of the reasons for using three stages in my 50 se amp,was that I trusted the saying: one more tubestage is better than a stepup trafo.
Hmm interesting results…
Well, the LL7903 performs really well in my system, except when I run some old CX-226 at lower anode current in the preamp so it doesn’t have the grunt and lose a bit of top end in the treble.
Otherwise, fantastic sound and with the benefit of galvanic isolation between the amp and the rest of the system.
I will have to do some listening tests with different drivers DHT and IHT…
Hi Ale,
Great comparison. Did you try with Zobel to get rid of those peaks? Using the LL1692A I could not see FR over 22khz ( sound card limit). Is it a way to blindly determine Zobel values ? Or at least approximate.
I like the idea of using the step up transformer for a complete isolation. What about the preamp driving a much larger capacitive load?
Best,
Radu
Ale,
In a previous discussion you mentioned the Audiophile M192 sound card. Are you able to measure FR beyond 22khz with this card?
Radu
Up to 96kHz which is very handy. FS is 192kHz and resolution is 24bit. I highly recommend building the Pete Millett’s interface.
Hello Ale,
having the need to catch up on some possible gain would be using the input to 4p1l lunhahl 7903, but using 1660 output connected to 4.5: 1, or it is preferable to a 1671 connected to 1: 1
Thanks for your help,
Greetings,
Andrea
Hi Andrea, it all depends on your following stage and the driving requirements. If it’s not a demanding load, then 1:1 should work fine
Ale
Thanks Ale,
the next stage is the input of a final singel ended, style Sakuma, with a valve 211 which drives through a transformer interstage another valve 211.
I have seen that would take about 6,5 Vrms input to the valve driver (211 tube) to obtain 15W output.
In the configuration of the preamplifier with 1660 Lundahl connected to 4.5: 1 can have only about 4 Vrms, but with the aid of a low output impedance.
The entrance of the valve 211 is perhaps the most challenging in the current?
I was wondering if the transformer would be more beneficial to use the entrance to the final, rather than the input of the preamplifier.
Thanks again for Your advice,
Greetings,
Andrea
4P1L phono pre?
1:32 LL1678
– 4p1l –
4:1 LL1671
LCR RIAA
– 4p1l –
1:2 LL1671
That would have almost enough gain for a MC.