V76s build - work in progress
- Thread starter Ilya
- Start date
Help Support GroupDIY Audio Forum:
dmp
Well-known member
Maybe I'm being thick but I missed it. Take the CMMI-10C. If you reproduce test circuit 1 or 2 from the datasheet, it should be flat out well beyond 20kHz
Now put that transformer into the v76, the secondary is connected to the grid, which has very high impedance. Somehow you are now getting -16dB at 20kHz? I don't get it.
I built mine with the Jensen 115k-E90 and it is fantastic. The Jensen 10k61-1M on the output. I'll see if I can find a frequency sweep from when I built it, or I can sweep it again this week and post.
No, you didn't. I mixed up forum threads. Here's the thread I'm talking about. Basically, the winding capacitance is the culprit of this roll-off.
I'm not talking about CMMI-10C. On the contrary, it's down only 0.5 dB at 20 kHz which is fine. With parallel primaries this increases to ~4 dB which is also tolerable (frankly, I should've skipped the lowZ option altogether).
-16dB happens with Sowter. Please read my post with graphs on the previous page once again and you will see that.
dmp
Well-known member
Ah, ok, lots to read through there. Interesting discussion
I misread the CMMI-10C graph - it does do well out to 20kHz in 1:10
Surprising info about the Sowter
I misread the CMMI-10C graph - it does do well out to 20kHz in 1:10
Surprising info about the Sowter
dmp
Well-known member
Thanks for taking your time to do the graphs.
I've finalized the build and here are some quirks that I found (and solved):
1. HF roll-off. As stated earlier, Sowter 1004 is not suitable for 200R input impedances. The solution was to switch to Cinemag 9994A (CMMI-10x will also suffice).
2. LF roll-off. This was buggering me as I was getting approximately -3dB at 20Hz which is not ideal. The solution was to increase the interstage cap. In the original schematic (V76S, 1:30 input tx) the coupling cap between V2 and V3 tubes is 35nF. This value proved to be too low and cut the LF too early. In contrast, in the V76M circuit (measuring version, 1:10 input tx) this cap is 10 times larger (0.3uF). And this is the correct value for extended LF response.
3. Oscillation. For some reason the input stage oscillated at approximately 300 kHz. I was able to solve this problem by inserting the resistor in series with the anode choke. But this solution didn't work for the second unit. So I put a small (100p) cap across the feedback resistors and this killed the oscillation. I removed the additional resistor from the anode choke and there was no trace of oscillation either. It's still a mystery to me as to where the HF "bleeds" and forms a positive feedback loop. I suspect that this is probably a layout issue. Anyway, I didn't measure any signal degradation with this 100p compensation cap, so I'm leaving it in circuit for time being.
4. My "virtual inductor" option for HPF (gyrator) proved to be a great and affordable solution. It worked as expected.
Here's the real inductor measurements (for all 3 switch positions):
And this is the gyrator measurements:
Now, I have a plan for next steps.
First, I'm going to write up a full building guide and put it on the site. All schematics will be available as well.
Second, I'm going to look into a scaled down version of this circuit. Since chokes are rather expensive components, I'm going to try to substitute them with CCS. Will see how this works out - need to do quite a bit of reading about this since I have not that much experience with tube circuits ATM. I'm going to strip this version of most of the options - I'll leave the tube DI option, pad and phantom. I might leave the HPF and high shelf in place, but this depends on the price of these options. My goal is to make bare bones V76 and see how much that would cost in the end.
If you guys have any input on the issues that I had during the build, I'd gladly discuss them.
dmp
Well-known member
Try shielding the chokes. I folded sheet metal into a can, wrapped it with adhesive backed mu-metal.
I am also wary of the transformer test results---the Sowter in particular is simply unusable. There's no way a company could charge 150 for a transformer that rolled off, especially considering the original V76 input is close to flat at 20k. Even giving the Sowter a pass b/c there are no formal specs listed, the Cinemag directly claims to achieve flat to 20k with primaries in parallel, 150 ohm source. Would be pretty bad false advertising. Was the impedance of the signal generator adjusted for each primary configuration?
I've managed to achieve sort of reasonable result with Sowter if I load its primaries with 50R. But this is not a viable solution, obviously, because it will load the mic too much. Besides, Sowter claims that their input transformer is for 200R load.
I'm not going to risk and try any other 1:30 trafos, because, as wise people here say, nobody makes good 1:30 mic transformers nowadays. I'm completely satisfied with Cinemag performance, and given that I'm using their products for many years and they always offered exceptional quality, I'm going to rely on Cinemags in my "stripped down" V76 version.
I still have the unit and transformers at hand, so I can run some more tests if you tell me exactly what you want me to test.
Is it possible that there is something about the V76 circuit that reduces the performance of these transformers? Looking at the schematic the input terminations are very complex compared to a standard tube preamp.
If you're talking about the caps in-between the primaries, these are for HF roll-off, not for termination.
is this rolloff constantly engaged? Is it in the audible range or just to tame some sort of ringing? This is very opaque to me; I've never seen another design like it.
Yes, this is not user switchable. If you look at the measurement versions of V76, there's no such caps between the primaries. I agree, this is not very obvious and I've never seen such solution anywhere.
I read somewhere that this is an HPF and there's another LPF downstream to limit the bandwidth for broadcast use. These filters aren't present in the V76m version.
I read somewhere that this is an HPF and there's another LPF downstream to limit the bandwidth for broadcast use. These filters aren't present in the V76m version.
I did some more digging, and I read the Cinemag datasheet wrong. They claim it is for 150/600 mics, but really best performance is as you described, 50 ohm load. Hidden in the datasheet, and even then -2db at 20k. The plots they show describe 150 and 600 ohm, but don't mention which circuit they used, leading me to believe that they were charted with primaries in series, 1:12 rather than the larger ratio. A little bit misleading if you ask me.
I‘m really wondering why it seems more difficult today to build linear high ratio transformers than 40-50 years ago although we have precision CNC winding machines and “better” materials today...
I have several vintage 1:20 or 1:30 mic input transformers here and all of them have great frequency response way beyond 20khz. The v76 input transformer measures flat to 50KHz with a source resistance of 200 ohms. The Sennheiser 1:30 I have here have a slight rise but can be made flat to 40KHz...
I have several vintage 1:20 or 1:30 mic input transformers here and all of them have great frequency response way beyond 20khz. The v76 input transformer measures flat to 50KHz with a source resistance of 200 ohms. The Sennheiser 1:30 I have here have a slight rise but can be made flat to 40KHz...
I’m pretty sure the cap between the primaries are for LF roll off not HF. It is to cut below 40Hz. There is a HF roll off in the circuit but that’s the BV517 choke after the selectable low cut filter. I think it filters out above 15Khz.
There are a few tube preamps with this arrangement of caps between the primaries. Some are even switchable. I think one of them is the “portable” V45.
Last edited:
Why are the transformers not rotated 90 degrees to each other and/or encapsulated with mu-metal?
Best regards?
jokeramik
