12AT7 Inverter in Poseidon Mark IIIs

[WLK]

I have a large supply of 12AT7s and I would like to use them as the inverter in place of the 12AU7 in the Poseidon MKIII boards.
Are there any disadvantages or advantages to using a 12AT7 as an inverter?
What value of plate resistors should I use for R10 & R11?
What value of current setting resistor should I use for R9?
Are there any other component values that I need to change?

[EWBrown]

IIRC, the VTA series of ST70 driver boards use 12AT7s in both VA and LTPI stages. I've used a couple of these boaards (from Roy Mottram) with veri nice results in ST70s.

From my immediate recall, the tail reistor is 12K and the two plate resistors were 51K. Standard ST70 B+ voltages apply. The combined plate currents were about 4.25 mA, so the 12K tail resistor sees around 51VDC across it, and the two plate resistors about 114VDC across each.

Since the Poseidon and ST70 "red board" use the CCS chip, the "tail" voltage is around 38V less, so the plate resistors could be increased to accomodate the extra approx 38VDC I'd go with 62K for the LTPI plates.

The CCS runs the total LTPI current around 6.8 to 7 mA, so each plate sees half of that, or 3.4 to 3.5 mA each.

The 12AT7 also makes a nce VA stage, and its plate and cathode resistors wil have o be changed to accomodate it.


THis is just from memory, so I could be a bit off, but it should be a good starter for you.


/ed B

[WLK]

EW:

Thanks for the information.

It is greatly appreciated and very timely since today I am planning on completing the rebuild and upgrade of a pair of Quicksilver monoblocks using Poseidon MKIII boards as the front end. I have been working on this project for about a week and have rebuilt the power supplies using solid state rectification and higher voltage rated filter caps, installed grounding buses, upgradesd the bias supplies and completely rewired the amps.

The Quicksilvers used 12FQ7s for the input voltage amp and the phase inverter and I was not very satified with the sound quality of the front end design or availability of these tubes. The amps always sounded very "dead" like too much feedback was being applied.

I would like to clarify that the current setting resistor R9 on the LM334 remains at 10 ohms and the two plate ressistors R10 and R11 should be around 62K in order to use a 12AT7 as the inverter?

Regards
WLK

[WLK]

EW:

I read you post once again more carefully and realized that your recommendations were based upon Stereo 70 voltages, which are significantly less than MKIII voltages.

The Quicksilver monoblocks that I am rebuilding and upgrading have a B+ voltage of about 490 VDC which is very similar to the MKIII. I have rebuilt the entire power supply with solid state rectification and more capacitance but as per the dropping resistors and voltages specified in the Poseidon / MKIII instructions.

I did some calculations based upon you recommendations and determined that the plate resiistors (R10 & R11) for a 12AT7 in the inverter position in the Poseidon boards at MKIII voltages should be about 100K ohms and the current set resistor on the LM334 (R9) should be 10 ohms.

This seems to be a common value for plate resistors for 12AT7 inverters running at similar voltages in other designs.

Shannon had instructed me when I bought the boards that If I used a 12AT7 as an inverter that the voltage at the base of the LM334 should be between 8 and 14 VDC. He indicated that I could accomplish this by changing the plate resistors or changing the current set resistor on the LM334. I was reluctant to experiment because I was unsure of a good starting point.

Based upon your recommendations, I now have a good starting point that I have a degree of confidence in and I will let you know how things work out.

Regards
WLK

[Ty_Bower]

I'm interested in the results, too. I have a boatload of 12AT7 tubes, and a mountain of 12AZ7 (which are supposed to have very similar characteristics to a 12AT7).

While I'm content to run a 12BH7 in the phase splitter socket of my Poseidon boards, I'm curious to see how the 12AT7 works out for you.

[EWBrown]

I should have read the original post closer (but then I was still in a power outage "haze" brought on by sleep deprivation)... I'm slowly recovering back to "normal" but that will take a few more days and some more decent uninterrupted nights' sleep.


My main 12AT7 experience is with ST70 drivers, and as the VA in various 2A3, 6B4G abd 300B SET designs, both Bottlehead, and my own "mutants" on BH's theme.

The 10 ohms resistor for the LM334 would set the total current to around 7 mA, so half that or 3.5 per 12AT7 section. THat should be fine, I've run 12AT7s in VA duty at 3.4 to 4 mA, and the ST70 LTPIs aer 4.5 mA or 2.25 mA per trioode sectioin, which seems low, but still works fine.

100K per plate resistor would give you around 350V drop across each plate resistor, so that value may be high, 62K may be low, I'd ain for something in between, perhaps 82K, 2W, for a good starting poing (that gives around 280 to 290V drop across each resistor)

Usually the second plate rsistor shopuld be around 5% to 10% higher, but it isn't absolutely necessary.

12AT7s are pretty rugged,, somewhat more so than 12AU7s. Even though the plate structure seems small, they still have plenty of "ballz" Yellow_Light_Colorz_PDT_06

I recently re-calculated the component values to run the "classic" VTL VA / LTPI circuit on 365V B+ in order to run PPP EL84s, in a yet to be resuscitaqted old project that I furtively started a few years ago.

Basically a "ripoff" of the VTL "Tiny Triode" Monoblocks, 45W in PPP pentode mode, and 26W in PPP triode mode. Tube CAD showed that I only had to change the LTPI plate resistors to 43 or 47K. The VA remains unchanged and the NFB loop R and C values are YTBD.

VTL runs their EL84s at almost 400V B+ and the plate current at 30 mA, fixed bias. I'll stick with the tried and proven ST35 standards of 365V and 35 mA, as I can use standard ST35 PSU design and parts.

/ed B

[WLK]

Ed

Thanks for the additional information and suggestion for the 82K plate resistors.

I am going to do a bit more work on the Quicksilvers this evening and hopefully finish them sometime tommorrow (Christmas shopping keeps getting in the way).

I will post my results tommorrow evening and maybe we can jointly determine the next steps to optimize the use of the 12AT7 as an inverter in the Poseiden MKIII boards.

Once again thanks for all the help.

Regards
Wayne

[EWBrown]

If you don't already have Tube CAD, you really should get it, it's very reasonably priced and very useful tool for trying out various designs before committing them to hardware. It covers most of the normal, and many not-so-normal amplifier configurations.

It does allow for adding more tubes, and that is a very easy process.


SE CAD is anotther very useful tool, and John Broskie also has Push Pull CAD (or Calculator) which I have not yet purchased.

SE CAD has a fixed tube "library" and does not allow for adding any more tubes, but it does allow adding output transformers quite easily.

/ed B

[Wiesiek Lipowski]

Why don't use use 12AT7 plus 6CG7 in the Poseidon project? Kara Chaffee - Audioxpress 2001, Mk III upgrade article - stated this combo goes together really marvelous. Shannon provides some info here regarding filament issue with 6CG7.

However, I use GE 5757 + RCA 12BH7A and am very pleased with the sound. I used to use the above pair before, but now my good old Mk III sound much better.

Good luck
Wiesiek

[Ty_Bower]

What a coincidence. I'm also using the GE 5751 and the RCA 12BH7 in my Mk3 amps.

[EWBrown]

I've been using 5751 and 12BH7A in my "deluxe" DIY ST35, and it has a nicer sound than the original 12AX7 and 12AU7 combinations tha tI've tried.

FWIW, the GE 6350 is another nice LTPI tube, heftier than either 12BH7 or 6CG7, but it has a "weird" pinout, similar to 5687 and 7044.

The 6350s are twin triode computer tubes, basically heftier versions of 5687s, and they work especially well in audio circuitry. VTL uses ths in their bigger power amps, as the LTPI tubes.

I ended up with a bunch of "pulled" 6350s (from the MIT computer lab) at a swapfest several years ago, and I still need to properly test them, beyond the initial emissions testing that I did last year. Tey all passed as "good" but that's only proof that they are not totally dead Yellow_Light_Colorz_PDT_11

The 6350s also have interesting possibilities as low power SET outputs, with their sections connected in parallel.

/ed B

[WLK]

There seems to be a fair amount of interest in this subject so I thought that I would post some quick preliminary results on trying to us a 12AT7 as a phase inverter in the Poseidon circuit.

I have rebuilt a pair of Quicksilver monoblocks with solid state rectifiers, 400 UF of B+ filtration and a feedback circuit that has been modified as per Shannon instructions for using Hammond 1650N output transformers.

All voltages are as per Poseidon MKIII specs.

I tried using 92K plate resistors for R10 & R11 but the plate voltages are way too low.

I installed 68 K plate resistors as per the the standard instructions for the MK III and tried both 12AU7 & 12AT7 tube types.

The 12AT7 yielded better results for my listening tastes with proper balance, good dynamics and highly defined bass, midrange and highs. Certainly an improvement over the original dead sounding Quicksilver front end using a 12FQ7 VA and a12BH7 phase inverter.

The 12AU7 sounded more like a stock Dynaco MKIII with mushy bass and subdued mids and highs.

This could be due to the chacteristics of the Quicksilver output transformers and the higher B+ filtration.

In the next few day I will install 56K plate resistors and try the test again.

I would like to increase the plate voltage on the 12AT7 to about 250 VDC as per other documented 12AT7 inverter designs.

I will also try lowering the current of 2.5 to 3.0 MA per plate.

I will post my results in a few days.

Regards
Wlk

[EWBrown]

Another consideration with 12AT7 in place of 12AU7 in the Poseidon circuit:

IIRC the LM334Z CCS ran the 12AU7 (and 6189) cathodes about 13V above ground, running approximately 3.4 mA per triode.

The CCS voltag will be substantially lower (around 4.25V) with a 12AT7, which could lead to some operational variances in the LTPI circuit.

In the VTL circuit, the LTPI plate resistors were 47 and 51K, the tail resistor 12K, with current set at 4.25 mA, whicch "floats" the conjoined cathiodes at 55V above ground, (51V across the 12K "tail" and another 4.25V across the shared 1K cathode resistor). The LM334Z CCS will probably be closer to the 4.25VDC above ground.

With the lower cathode voltage, the plate resistors should drop around half the B+ voltage, for 400VDC LTPI B+, and assuming 2.1 mA per section, the plate loads would be 200V / 2.1 mA or approx 91K.

This 2.1 mA is rather low plate current for a 12AT7, and tube CAD gives all sorts of warnings about this "suboptimal" plate current, which could lead to increased distortion - though I never noticved this in the modified VTA/VTL ST70s

In the "real" world, the 12AT7 would probably be better off with a minimum of 3.5 to 4 mA per section, so aim for 51 to 56K per plate resistor. Typically the "right hand" plate resistor shoule be about 4K to 5K higher resistance, in order to maintain better AC (signal voltage) balance.

Sine I don't have any actual "hardware" too test this on, I can only use tube CAD and plate curves for now...

/ed B

[Shannon Parks]

The CCS operating point is probably the main consideration - thanks to to Ed for bringing it up. Forgetting this point and not having TubeCad'd it I figured, "why not just drop it in?". I would like to see greater than 7V on pins 3 and 8 at a minimum. Stop operation immediately if lower. Then adjust the current set resistor to get a voltage in the 9V-13V range.

[WLK]

Ed and Shannon

Thanks for the additional information and suggestions.

I was called away for a few days but tommorrow morning I plan on doing some more work on this project.

As per your suggestions I will first install two new 56K plate resistors and run the amps to determine what the voltage is at the combined cathodes.

I will then adjust the current set resistor so I get a voltage of between 9 and 13 volts at the combined cathodes.

What is the concensus on the requirement for using unbalanced plate resistor?

If unbalanced plate resistors are the favored solution, should they be 52K & 56K or 56K & 60K?

Which location (R10 or R11) uses the higher value plate resistors ?

I will report the test results tommorrow evening.

Once again thanks for all the help.

Regards
Wlk