Tube phonostage RIAA design

[emi2345]

Hey everyone, many thanks for all your advice.

After doing a bit more noodling around and learning about CCS, I think that I may be able to get enough gain using two 6021 stages with a CCS as the anode load for each stage. A CCS anode load can bring the gain up to nearly the mu of the valve, and with a mu of 35 two 6021 stages would provide 60dB of voltage gain. 30dB per stage, 60dB gain overall, 20dB loss for the passive RIAA leaves the target 40dB.

The advantage of the 6021 vs the 6948 (aka 6112) is the higher gm of 5.4mA/V and lower anode resistance of 6k5 ohms. This should mean lower noise from the stage itself and also lower resistor values for the RIAA network. If I parallel up two triodes for the input stage, I should get a anode resistance of around 3k3 so I could use a 33k build out resistor for the RIAA network.

Given that I now don't believe I can implement a differential tube amplifier stage that has BOTH perfect balance from the CCS in the cathode, and maximum linearity and gain from a CCS as the anode load, I think I will stick with a single-ended design for now.

Operating point wise, the most linear part of the 6021 I can find on the curves is 5ma with 2v bias, 123v on the anode, 0.62W dissipation (88%). If I use a 250v supply that would make 123v on the anode half the supply voltage which is handy for a DC coupled cathode follower. I will use the same operating point for each stage.

I get different absolute max values for Va on different datasheets, I have the standard GE non-WA 6021 which quotes 160v as the design maximum but in the tables they give examples using a 225v supply so will I be ok with 250v at turn-on?

1. Parallel 6021 triodes with a cascoded mosfet CCS (perhaps DN2540?) anode load. per triode: V supply 250v, 5mA, 2v bias, Va 123v Pa 0.62W (88%). LED bias. mu = 35 = 31dB; Rp = 6k5/2 = 3k3

2. RIAA network starting with 33k series resistor, other values calculated using http://www.mh-audio.nl/Calculators/CalculateRIAA.html

2. 6021 common cathode gain stage with cascoded mosfet CCS anode load : Vsupply 250v, 5mA, 2v LED bias, Va 123v Pa 0.62W (88%)

4. 6021 DC-coupled cathode follower: Vsupply = Va = 250v, Vg 123v, 5mA, CCS cathode load, arc protection.

Many thanks

 

[ruffrecords]

You overlooked word entire.
This ratio (squared Ia/gm) for entire range for 12AX7 is more than 16.
In a usual range of Ia+Ig2 of 10 to 25mA for D3a, this ratio is below 2,
unlike in usual range for Ia=0,5 to 1,5mA for 12AX7 this ratio is above 4.

This is in line with what Merlin Blencowe told you seven years ago on diyaudio.com:
2,7225 isn't small error for you but difference in gm of more than 20 may (or may not?) make a difference. Interesting.

Indeed it is interesting. Merlin and  have had a long and fruitful discussion on this topic, not least about the various ways in which people choose to measure noise and it effect on the values obtained. However, that was in the days before REW and the ability to see the actual noise spectrum of a preamp. What I will do is buy myself a D3a or two and build it into a version of my current mic pre design and measure its noise to see if in practice there is any improvement.

Cheers

Ian

 

[ruffrecords]

I have purchased a couple of D3a tubes and spent the evening with LTspice working out how to integrate the D3a into my standard mic pre design (the result is attached). With a 1:10 mic input transformer and a 2:1 output transformer this should be able to offer a gain range from 60dB down to 26dB. I will design a PCB, get some made and try it out. If it works OK I am happy to pass on the reaming boards to others to play with. And of course I will see how good its noise performance is too.

Cheers

Ian

 

[gyraf]

Looks good indeed?

Can you get D3a at decent prices, and what about the sockets for it?

/Jakob E.

 

[ruffrecords]

Looks good indeed?

Can you get D3a at decent prices, and what about the sockets for it?

/Jakob E.

I found a couple on eBay for around £20 so not too bad. One is made by Siemens and the other is Telefunken. Socket is no problem as it has a B9A base. D3a heater current is the same as the ECC88 so I plan to wire their heaters in series and run them from 12V.

Cheers

Ian

 

[gyraf]

Thanks Ian!

For a moment I confused it with the C3g that it's hard find, and to find sockets for (at decent rates anyway).

The D3a/E810F is a very nice tube for this sort of input stage, following topic from side line...

/Jakob E.

 

[Jonte Knif]

My experience with D3a is that microphonics is usually way too high for low level signals. Perhaps one can find good samples, I haven't found.
Same story with C3g. Way too microphonic for anything else than line level signals usually. 

This is to be expected. High Gm means grids very close to cathode and therefore even the smallest movements become problematic. But there are differences for sure. The last high Gm tube I tested was E180f, and it was superb in line level use, so might actually be good for RIAA. It is also fairly cheap.

 

[ruffrecords]

This is to be expected. High Gm means grids very close to cathode and therefore even the smallest movements become problematic. But there are differences for sure. The last high Gm tube I tested was E180f, and it was superb in line level use, so might actually be good for RIAA. It is also fairly cheap.

I have not delved too much into the relationship between the mechanics of a tube and it parameters but I thought closeness to the cathode relative to the anode was the prime determinant of mu and that gm is primarily determined by the number of turns per unit length of the grid. Of course I may be wrong.

Cheers

Ian

 

[Jonte Knif]

Yeah, you may be right. I've taken apart all tube types which I use and it has just been my observation that the grids are very close to cathode in these tubes and certainly the grids are super tightly wound too. E180F has a grid so tight that with bare eye it looks like a film, individual wires are impossible to see.

 

[ruffrecords]

Yeah, you may be right. I've taken apart all tube types which I use and it has just been my observation that the grids are very close to cathode in these tubes and certainly the grids are super tightly wound too. E180F has a grid so tight that with bare eye it looks like a film, individual wires are impossible to see.

Of course, the D2a is both a high gm AND high mu tube and it has certainly been my experience that hi mu tubes tend to be more microphonic. I spent a lot of time tapping various versions of the 12AX7 until I settled on the Sovtek 12AX7WA as being the least microphonic of them all, NOS included.

Cheers

Ian

 

[merlin]

I've never tested a true D3a but the 6J52P is similar and is indeed very quiet over a range of currents. But yes, it is very sensitive to microphonics. High-gm goes hand in hand with microphonics -those frame grids are as fine as spiders webs! I've never had a problem with 12AX7 microphony, only the occasional bad apple, which could happen to any tube type. (The most microphonic tube I have ever had was a bad 12AT7!)

 

[gyraf]

..couldn't we just servo our way out of the microphonics, I mean with a fast-acting mechanical arrangement like the gimbals for video cameras?

 

[ruffrecords]

..couldn't we just servo our way out of the microphonics, I mean with a fast-acting mechanical arrangement like the gimbals for video cameras?

I am sure I have seen at least one ancient product where the first stage tube socket was mounted in rubber? suspension to mitigate microphonics. I looked at this some years ago when searching for low microphonic tubes. I found some  computer flexible mountings that I was considering using. I also seem to remember at some time seeing a rubber grommet used for the same purpose. But these were all point to point solutions. I am not sure how to apply this to PCBs. Maybe you could but grommets in each of the PCB mounting holes. Having a nice large mass like and output transformer on the PCB would certainly damp down the transmission of any high frequency knocks.

Cheers

Ian

Cheers

Ian

 

[JohnRoberts]

I am sure I have seen at least one ancient product where the first stage tube socket was mounted in rubber? suspension to mitigate microphonics. I looked at this some years ago when searching for low microphonic tubes. I found some  computer flexible mountings that I was considering using. I also seem to remember at some time seeing a rubber grommet used for the same purpose. But these were all point to point solutions. I am not sure how to apply this to PCBs. Maybe you could but grommets in each of the PCB mounting holes. Having a nice large mass like and output transformer on the PCB would certainly damp down the transmission of any high frequency knocks.

Cheers

Ian

Cheers

Ian

I'm not sure how it would reduce sound conduction but I remember an odd tube sub chassis that was hanging suspended from two pointed screws... I guess this would allow the assembly to swing freely side to side resisting lateral shaking.. or not... This was in an old Western Electric box from the 40s maybe 30s?

JR

 

[gyraf]

frame-grid tubes are often much more sensitive to mechanical vibration from one particular angle, perhaps that would make sense?

 

[merlin]

I am sure I have seen at least one ancient product where the first stage tube socket was mounted in rubber?

The one on the left is from my book -a socket mounted on a rubber baffle. It came out of a cine projector.
The right image is a floating sub-chassis suggested in Morgan Jones' book. I've seen a variation on the same theme where the sub-chassis was mounted with screws and soft springs instead of elastic.

 

[moamps]

The problem with the elastic suspension of the tube base is that when connecting wires and other elements to the base, this suspension stiffens. Therefore, it is a better option IMO to suspend the entire circuit (pcb or p2p) in the housing. I got much better results and mechanical insulation that way. Of course, the tubes should be placed in sockets with a shield top mount.

 

[ruffrecords]

The problem with the elastic suspension of the tube base is that when connecting wires and other elements to the base, this suspension stiffens. Therefore, it is a better option IMO to suspend the entire circuit (pcb or p2p) in the housing. I got much better results and mechanical insulation that way. Of course, the tubes should be placed in sockets with a shield top mount.

I have never been able to find PCB mounting sockets that take shields.

Cheers

Ian

 

[moamps]

I have never been able to find PCB mounting sockets that take shields.

When I use a pcb, I usually use shielded tube socket intended for p2p and chassis mounting, I mount it on pcb and connect the pins from the socket to pcb with short wires.
Have you seen these sockets?
https://www.banzaimusic.com/tube-socket-pc-9-shb.html
https://www.banzaimusic.com/tube-socket-noval-pc-xt.html

 

[ruffrecords]

When I use a pcb, I usually use shielded tube socket intended for p2p and chassis mounting, I mount it on pcb and connect the pins from the socket to pcb with short wires.
Have you seen these sockets?
https://www.banzaimusic.com/tube-socket-pc-9-shb.html
https://www.banzaimusic.com/tube-socket-noval-pc-xt.html

I have some very similar to both those types. I bought them a couple of years back when I was asked about a ruggedised mobile tube mixer. Unfortunately it is almost impossible to fit a current production Russian tube into these sockets. It takes a hell of an effort to get the tube into the socket and when I try to remove it you have to apply so much pressure I am afraid I might break the tube and cut my hand to ribbons (I have just been over to my workshop to try this out). Interestingly, the two NOS D3A tubes I just purchased, the stiff, do go in and come out much more readily. Perhaps current production tube pins are slightly larger in diameter so In think I might use these for this project. I might try some NOS PCC88 tubes instead of the current production 6922 tubes I normally use for the output stage for the same reason.

I don't know if the Banzai ones are any different to the ones I have (they don't look quite the same) so they may not have this problem. So I will buy a couple to try out.

Cheers

ian