I mounted the stock values of RIAA components on the bottom of the board, then used the headers shown with which to parallel in additional C or R. You can also see the paralleled resistor to get the first stage CCS@20mA:
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
HV and Filament supplies on a new and separate compartment -- it will be shielded to protect the innocent and unwary:
An externally hosted image should be here but it was not working when we last tested it.
Hi Guys,
Including some mods here is my version of this fantastic design:
(In Dutch, but at least the pictures are clear for everyone)
Arno
MC Element met SUT upgrade
Including some mods here is my version of this fantastic design:
(In Dutch, but at least the pictures are clear for everyone)
Arno
MC Element met SUT upgrade
Excellent article SY, thanks.
It got me thinking about the possiblity of an MM version, as mentioned a few pages back. Having done a few back-of-an-envelope calculations, I reckon that replacing the D3a with another ECC88/6DJ8 gain stage (identical to the second stage) would work quite nicely.
The equivalent noise resistance is 300 Ohms, giving a 1st stage SNR (assuming no transformer and a 100 Ohm grid stopper) of -79dB, cf. the -81dB figure that you get from a 630 Ohm cartridge.
Adding in a Sowter 1:1 600-600 Ohm transformer (1.9dB insertion loss, 150 Ohm secondary resistance) gives a SNR of -76dB, but no doubt a significant common-mode improvment.
The input capacitance works out to be about 56pF (including the 5pF for strays) although that number varies by 2.7pF between sections of the 6DJ8 - probably better to use the section on pins 6,7, and 8 for the input on both channels for the lower value. I have no experience of de-ringing transformers, but I'd expect that to be low enough to get decent results.
The 1st stage output signal level is about 165mV, and the output impedance is about 2K6 Ohms - perhaps different enough from the D3a's to make re-calculating the RIAA to R1 = 27k worthwhile, to keep the dependance on the tube's Ra down.
The output with the 5mV output of many MM carts would be ~0.54V at 1kHz.
Other possible tubes are the 6DX4 (Ra = 2K7, Mu = 30), or the 6DY4A (Ra = 2K5, Mu = 28) - but they seem to be on the rare side. I haven't looked into the input capacitance or noise resistance, partly because datasheets are thin on the ground.
Does anyone have any immediate thoughts?
Specifically, I'm unsure as to whether the 1:1 transformer should be of the 600 Ohm or 10K variety. My instinct tells me that the lower-impedance variety is the one to go for, as you might want to keep the winding impedances around the value of the cartridge imppedances - but OTOH, the cartridge load is 47K, so should it be the higher-impedance version. TBH, I'd probably build without the input transformer to start with, and then add one as funds become available.
It got me thinking about the possiblity of an MM version, as mentioned a few pages back. Having done a few back-of-an-envelope calculations, I reckon that replacing the D3a with another ECC88/6DJ8 gain stage (identical to the second stage) would work quite nicely.
The equivalent noise resistance is 300 Ohms, giving a 1st stage SNR (assuming no transformer and a 100 Ohm grid stopper) of -79dB, cf. the -81dB figure that you get from a 630 Ohm cartridge.
Adding in a Sowter 1:1 600-600 Ohm transformer (1.9dB insertion loss, 150 Ohm secondary resistance) gives a SNR of -76dB, but no doubt a significant common-mode improvment.
The input capacitance works out to be about 56pF (including the 5pF for strays) although that number varies by 2.7pF between sections of the 6DJ8 - probably better to use the section on pins 6,7, and 8 for the input on both channels for the lower value. I have no experience of de-ringing transformers, but I'd expect that to be low enough to get decent results.
The 1st stage output signal level is about 165mV, and the output impedance is about 2K6 Ohms - perhaps different enough from the D3a's to make re-calculating the RIAA to R1 = 27k worthwhile, to keep the dependance on the tube's Ra down.
The output with the 5mV output of many MM carts would be ~0.54V at 1kHz.
Other possible tubes are the 6DX4 (Ra = 2K7, Mu = 30), or the 6DY4A (Ra = 2K5, Mu = 28) - but they seem to be on the rare side. I haven't looked into the input capacitance or noise resistance, partly because datasheets are thin on the ground.
Does anyone have any immediate thoughts?
Specifically, I'm unsure as to whether the 1:1 transformer should be of the 600 Ohm or 10K variety. My instinct tells me that the lower-impedance variety is the one to go for, as you might want to keep the winding impedances around the value of the cartridge imppedances - but OTOH, the cartridge load is 47K, so should it be the higher-impedance version. TBH, I'd probably build without the input transformer to start with, and then add one as funds become available.
I think investigating pentode connected D3A might be worthwhile, certainly miller capacitance no longer be an issue, partition noise doesn't seem to be an issue with this type or the 7788 in my very limited experience with phono front-ends.
I don't recommend attempting to use a CCS load on a pentode, but the CCS could be reconfigured to drive a shunt regulator which could provide a low noise low source impedance dc supply for the plate load resistor. The plate load resistor will ~ define the source impedance driving the next stage.
D3A noise resistance pentode connected is about 120 ohms IIRC, still rather good..
In terms of transformers you don't need them with a MM type, their primary purpose is voltage gain, although galvanic isolation/balanced operation is an added benefit. (Noting that I use neither feature in my system with LOMC cartridge.)
You would need a 50K:50K transformer in this application, and frankly unless you were willing to spend a large sum of money this might prove an unnecessary compromise.
I don't recommend attempting to use a CCS load on a pentode, but the CCS could be reconfigured to drive a shunt regulator which could provide a low noise low source impedance dc supply for the plate load resistor. The plate load resistor will ~ define the source impedance driving the next stage.
D3A noise resistance pentode connected is about 120 ohms IIRC, still rather good..
In terms of transformers you don't need them with a MM type, their primary purpose is voltage gain, although galvanic isolation/balanced operation is an added benefit. (Noting that I use neither feature in my system with LOMC cartridge.)
You would need a 50K:50K transformer in this application, and frankly unless you were willing to spend a large sum of money this might prove an unnecessary compromise.
I thought about this and I thought about what it takes to make a really good 50k:50k transformer, and I leave that path to others.
As a pentode, the D3a still gives pretty excellent performance in the first stage (the distortion spectra at these small signals are pretty clean), and if I were to be willing to give up balanced, that would be a high priority option. I'm not, so I've gone another route for MM...
D3a in pentode will give a wad of gain, easy to squeeze out 50-60dB.
As a pentode, the D3a still gives pretty excellent performance in the first stage (the distortion spectra at these small signals are pretty clean), and if I were to be willing to give up balanced, that would be a high priority option. I'm not, so I've gone another route for MM...
D3a in pentode will give a wad of gain, easy to squeeze out 50-60dB.
The interesting thing is that because the signal levels are so low it is possible to use silly small plate load resistors and get whatever gain you desire. As low as 1K is feasible with the D3A and 600 ohms with the 7788, gains comparable to triode connection are possible with no miller capacitance.
I have run MM and HOMC in a pinch into D3A triode connected despite the high miller capacitance with usually acceptable if not quite brilliant performance. Not that I recommend this..
I have run MM and HOMC in a pinch into D3A triode connected despite the high miller capacitance with usually acceptable if not quite brilliant performance. Not that I recommend this..
Thanks for the feedback guys. I suppose the 50k:50k transformer value makes sense, although as you say it would be difficult in practice.
"so I've gone another route for MM..."
You've got me intrigued now. I suppose that one brute-force method would be to throw more active devices at it - maybe a balanced-input cathode follower, with a balanced-to-single ended transformer between that and the first gain stage.
"so I've gone another route for MM..."
You've got me intrigued now.
I suppose that one brute-force method would be to throw more active devices at it - maybe a balanced-input cathode follower, with a balanced-to-single ended transformer between that and the first gain stage.
My original idea was to use p-FETs as a source follower, directly coupled via an LED into the cathode of the input tube (run grounded grid). Bench testing showed great results with a 12AX7. Problem is, no one is making good low noise p-FETs anymore, and the number of fakes is alarming. So it's not something I can put into an article for other people to build.
For MM, I ended up going in a totally different direction, and as a nice bonus, it looks to be pretty low cost - when it's finalized, I'll probably put it up on my website.
For MM, I ended up going in a totally different direction, and as a nice bonus, it looks to be pretty low cost - when it's finalized, I'll probably put it up on my website.