JTM 45 negative feedback with Mercury 'Radiospares Deluxe' OT

[loudthud]

I was thinking the exact same thing. The original amp using a transformer that this MM is supposed to be a "clone" of was 16 ohms with the tap from 16 ohms. And, according to loudthud, the 16 ohm arrangement "is not ideal". But is THAT the way the original was wired? If so, and since Mike is chasing original mojo with so much of this effort, just build the thing as a frikin' jtm45!?!

What I meant by "not ideal" is that the A-F winding is connected on top of the series connected D-E and B-C sections. I simply re-arranged the series sequence so the A-F section is always on the ground side, F is always grounded and A is always where the feedback is connected. That makes it a whole lot easier to do the switching. I'm working on the diagrams for a more complete post.

[g1]

No, 8k is strongly mismatched. See my previous posts. Max. power is achieved with Raa between 4k and 5k. Please check data sheets (loadline construction). Actually the KT66 is quite similar to the 6L6.

Ok, I guess you meant your posts in the other thread here: http://music-electronics-forum.com/t46603/
Thanks. I had never considered that the JTM's were using a technically 'wrong' transformer. All the replacements that I've seen for JTM45's using a pair of KT66 have 8K primary and I had never looked at the numbers, I just figured that 8K was 'matched'.
Plugging them into the bmamps calculator gives similar power figures to what you said (higher power for 4K primary).
I guess a practical take away is that anyone who wants more power out of their JTM45 can cut their load impedance in half for power gain. (assuming the OT is rated for the higher power)

[Chuck H]

There are reports that the 8k primary intended for a 16k load was indeed played into a pair of 16R speakers wired at 8 ohms (so, 4k primary). Then others that report that the 8k primary wired on the RS OT was matched for that impedance at 8 ohms. WTFK is my take so far. The available info is a total kludge.

[loudthud]

The Ideal OT would have a single winding with taps for 4, 8, and 16 Ohms. The taps would be at 50%, 70.7% and 100%. The Radio Spares OT uses series and parallel connections to it's split secondary windings. If they were simply connected is series, a 1P3T switch could be used but the wire in the transformer might be stressed by too much current on the 4 and 8 Ohm taps unless the power is dialed down to 25 Watts or so. Note that the RS OT has a slight error on the 8 Ohm tap at 75%. This equates to 9 Ohms so when an 8 Ohm load is connected to the 8 Ohm tap, the primary impedance falls to 7.111K.



The RS OT secondaries have been re-arranged so that F is always grounded and feedback is always taken from A. Some purists may object to this, but I challenge them to hear any difference. When coming up with these switching arrangements, you need all the help you can get.



The diagram I posted in reply 24 above has the disadvantage that if the switches are not set correctly, the wrong windings could be connected in parallel and that could damage the OT. Think of it this way. A-F is a 14V winding. B-C and D-E are 7V windings. In the 8 Ohm mode B-C and D-E are connected in parallel. Those parallel connected windings need to be connected in series with the A-F winding, but a wrong switch setting could connect them in parallel with the A-F winding.

Notice that switching between various impedances requires switching windings between series and parallel arrangements. There are a couple of ways to do this. In Example 1 you see that when the switch is in the lower position, you have two switch sections in series. A simple re-arrangement of the wiring puts those two switch sections in parallel (Alternate Example 1). Example 2 is another way to accomplish series-parallel switching but I used the Alt 1 example.



Using a single toggle switch, you can select between two impedances. With a DPDT switch you can select between 4 and 16 Ohms or 8 and 16 Ohms. Unfortunately it takes a 4PDT switch to select between 4 and 8 Ohms. Lots of old Vox amps have only 8 and 16 Ohm taps on their OTs.



Finally, if you can find a Four Pole Three Position switch, you can select between 4, 8 and 16 Ohms. The switch needs to be able to stand about 7 Amps for a 45 Watt amp making square waves. Don't be tempted to use those 1 Amp switches that are commonly available. They will just burn up. Ever notice how the impedance switch on old Marshalls is almost always burned up?

[Chuck H]

Yep! For ease of use, since I often use the old style Hammond 16xx OT's, my amps run 4 or 8 ohms (which is easiest with THAT OT secondary). Easy to do with a DPDT switch. Which is easy to find in the necessary current requirement. Big kudos for working out a multi impedance diagram for this OT (which has never been done as far as I can tell). But the "purists" will, of course, argue that anything "different" is heresy and will speculate in pseudo science about tonal anomalies WRT your diagram. So what? But who knows? If the switch is available people may be more interested in the advantage

[J M Fahey]

Yep! For ease of use, since I often use the old style Hammond 16xx OT's, my amps run 4 or 8 ohms (which is easiest with THAT OT secondary). Easy to do with a DPDT switch.

Can you please be more specific?
Googled Hammond 16xx OT and Angela alone has some 25 different ones.

[Helmholtz]

Ok, I guess you meant your posts in the other thread here: JTM 45 build 400v plate voltage, how to calculate OT primary impedance?

Seems I mixed up similar threads, sorry.

The output power results from the bmamps calculator are unrealistic i.e. way to high. At least if you use no-signal (idling) voltages. The results should be better with plate and screen voltages measured at full output power. Also the OT eats a few watts.

The power values I gave are real measured ones.

[Chuck H]

Can you please be more specific?
Googled Hammond 16xx OT and Angela alone has some 25 different ones.

Here's the diagram for the 16xx wiring. You can see that it's straight forward switching between 4 ohms and 8 ohms. I just omitted the 16 ohm option. The "easy wire secondary" version wasn't available when I started with these transformers and I never tried them because I didn't want to introduce an unknown element to an existing design since I don't build many amps anyway.

[Leo_Gnardo]

Here's the diagram for the 16xx wiring. You can see that it's straight forward switching between 4 ohms and 8 ohms. I just omitted the 16 ohm option.

I did this in a 120+ watt bass amp using the Hammond OT. The quandary "what to do?" disappeared once I decided the chance of mister Bass man using a 16 ohm cab was virtually nil. Also, Antique/CE offer a special switch set up to select 4, 8 or 16. Once I had the switch in hand, saw how flimsy the contacts were, and gave up the idea of installing it. Maybe in a low power amp, definitely not in a 4xKT88 behemoth.

[g1]

The output power results from the bmamps calculator are unrealistic i.e. way to high. At least if you use no-signal (idling) voltages. The results should be better with plate and screen voltages measured at full output power.

I'll pass that on to nick for ya.

[nickb]

Seems I mixed up similar threads, sorry.

The output power results from the bmamps calculator are unrealistic i.e. way to high.
The power values I gave are real measured ones.

I'd have to agree with you as I'd noticed it seems high too. I keep meaning to track down where it came off the rails but haven't had the inclination. Maybe I need to spend less time on MEF

Update:
I just did a quick cross check for a couple of EL34's at 430V into 3k. It seems the unexpected higher output is primarily attributable to the fact the screen grid is held as the the voltage specified in the inputs. Typically, in real life, the screen grids are fed via a resistor. In this case a 1K resistor will drop the screen grids to 360V or so, that's quite a difference. BTW, I assumed 95% transformer efficiency.

The tool was originally intended just to allow you to get plate curves for all those cases that you don't find in the average data sheet but it sort of morphed into more that that. Maybe I'll add a line for the screen grid resistance. One day...

Anyway, I don't want to derail the thread any more so I'll leave it there and let everyone get back to the matter in hand.

[Helmholtz]

I'll pass that on to nick for ya.

??

[Helmholtz]

I'd have to agree with you as I'd noticed it seems high too. I keep meaning to track down where it came off the rails but haven't had the inclination. Maybe I need to spend less time on MEF

Update:
I just did a quick cross check for a couple of EL34's at 430V into 3k. It seems the unexpected higher output is primarily attributable to the fact the screen grid is held as the the voltage specified in the inputs. Typically, in real life, the screen grids are fed via a resistor. In this case a 1K resistor will drop the screen grids to 360V or so, that's quite a difference. BTW, I assumed 95% transformer efficiency.

The tool was originally intended just to allow you to get plate curves for all those cases that you don't find in the average data sheet but it sort of morphed into more that that. Maybe I'll add a line for the screen grid resistance. One day...

Anyway, I don't want to derail the thread any more so I'll leave it there and let everyone get back to the matter in hand.

Thanks for your feedback. Didn't know that it is your calculator. Actually used it for the first time. Looks like a great, useful tool and I hope you find a way to fix the power issue. 95% OT efficiency may be a little optimistic, but I have no real values. Tube rectifiers increase sag and many KT66 actually show more residual DC plate voltage at Vg=0 (in other words have a higher DC on-resistance) than the datasheets suggests.

BTW, the KT66s in (many) original JTM 45s had individual 470 Ohm screen resistors plus a shared 1k in series. This corresponds to an equivalent of 2.47k per screen.

[Helmholtz]

There are reports that the 8k primary intended for a 16k load was indeed played into a pair of 16R speakers wired at 8 ohms (so, 4k primary). Then others that report that the 8k primary wired on the RS OT was matched for that impedance at 8 ohms. WTFK is my take so far. The available info is a total kludge.

Until solid evidence I tend to doubt that Marshall used the RS type OT in original Bluesbreaker combos. The reasons are 1)The KT66 equipped combos were produced only in the end of the JTM45 era ('65/'66), when most JTM45 tops already used the 8k Drakes. 2) The original schematic of models 1961/1962 shows an OT with a standard multi-tapped secondary. 3) The OT in the schematic has a 100V secondary winding, which only the Drakes had.

[pdf64]

...The tool was originally intended just to allow you to get plate curves for all those cases that you don't find in the average data sheet but it sort of morphed into more that that. Maybe I'll add a line for the screen grid resistance. One day...

I think user competence has to play a part in these things, it's surely up to them to plug in the Vdc applicable to the relevant conditions? The vintage tube info doesn't spoon feed so no obvious need for your tool to do so.
Sure it would be nice to have it all singing and dancing, but it may then need to include HT sag for both the plate and g2 nodes (even bias sag too!), along with local ac nfb caused by the individual / shared g2 resistors, and the whole thing would be in danger of becoming unfeasibly complex for both the developer and the noob user.

Until solid evidence I tend to doubt that Marshall used the RS type OT in original Bluesbreaker combos. The reasons are 1)The KT66 equipped combos were produced only in the end of the JTM45 era ('65/'66), when most JTM45 tops already used the 8k Drakes. 2) The original schematic of models 1961/1962 shows an OT with a standard multi-tapped secondary. 3) The OT in the schematic has a 100V secondary winding, which only the Drakes had.

FWIW the Marstran db for early Marshall lists a 64/65 2x12 1962/T with a RS OT http://marstran.com/Historic%20Data%20Base.htm