Ec Tremolux design musings

[tubeswell]

Had an EC Tremolux in for a service the day before yesterday. Had a 'funny bassy-subharmonic distortion like cone cry' when dimed. There was blocking distortion evident. (The 1k5 grid stoppers don't do anything). Had little EHX 6V6Gs in it, which were running like crazy - V4 was idling at 10.7W, V5 at 12.3W. Tried dialing in a higher bias voltage (only 38VDC available with the crazy bias supply design). The result was the plate voltages went up to ~440ish and the dissipation was still at 10.5W for V5. The bias supply voltage completely disappears when the standby switch is off (inherent design flaw).

With JJ6V6S, it at least biases up properly. The plates are sitting at ~410V@26.1mA (10.6W), which is okay for a 14W tube, and the customer is very happy with the sound.

I think a 'better' improvement would be to move the standby switch to the B+ rail (or remove it altogether) and have a BFPR type of bias supply in this amp. The way it is in the stock fender configuration, its difficult to get a more negative bias voltage if you need it. After 38VDC there is nowhere to go, and the B+ is quite high.

Couldn't find a decent schematic, so I traced one up for kicks and measured all the values etc.

[Leo_Gnardo]

When I see "EC Tremolux" I think, oh the one that's on the Blind Faith cover, a brownie with 2x10 cab. Not a tweed.

Looks like Fender has come up with yet another pile of piss, doubt Mr. Clapton would actually stoop to play such a defective thing. How do they get away associating it with him?

Thanks for your post tubeswell, if I ever run across one now I'll know what to do.

[Jazz P Bass]

That's the amp with the 12 amp heater fuse.

[tubeswell]

That's the amp with the 12 amp heater fuse.

I didn't notice. The schematic in the service manual that I've just obtained says 1 x 4A for each of the 6V and 5V windings

[Justin Thomas]

"Considered to be serviceable to the component level..."

Well, THAT is a refreshing change!!!

Justin

[Enzo]

I suspect that refers to warranty work, and is the opposite of the "do not repair" list. Translate: "We'll pay you to do component repairs on this model."

[Justin Thomas]

I worked on a Vibro-King last month, and was surprised to see a black fiber eyelet board, indeed printed in white ink with component designations. I'm guessing this is similar. I thought it was the best of both worlds, because it is indeed a "printed" circuit board, but without traces. I enjoyed working on it. No traces for my ham-handed removal techniques! Maybe if I do another amp with a turret board, I'll try something similar...

Justin

[66 Kicks]

Those two 0.1uf capacitors between the winding and the bias supply seem to be way too small for the application. Any chance those are actually 1.0uf?

You can raise the absolute value of the bias by increasing the value of those two capacitors. If they are indeed 1.0uf, replacing both of them with a single 1.0uf @ 630V capacitor will probably almost double the absolute value of the bias.

They are probably using two capacitors in series for the same reason that Vox used two. The transformer unleashes everything it has across those capacitors when the power is first turned on and a single 450V capacitor doesn't have a sufficient voltage rating for this. Vox used balancing resistors for those capacitors so that each one would take an equal load.

That bias supply will drain down on standby, but it should charge up really fast with no resistors in the charging path.

[Leo_Gnardo]

Those two 0.1uf capacitors between the winding and the bias supply seem to be way too small for the application. Any chance those are actually 1.0uf?

You can raise the absolute value of the bias by increasing the value of those two capacitors. If they are indeed 1.0uf, replacing both of them with a single 1.0uf @ 630V capacitor will probably almost double the absolute value of the bias.

They are probably using two capacitors in series for the same reason that Vox used two. The transformer unleashes everything it has across those capacitors when the power is first turned on and a single 450V capacitor doesn't have a sufficient voltage rating for this. Vox used balancing resistors for those capacitors so that each one would take an equal load.

That bias supply will drain down on standby, but it should charge up really fast with no resistors in the charging path.

Ampeg used a similar technique in their V2, V4, VT40, VT22 series, also found in Sundown amps. Marshall 900 do it too, deriving a bias supply thru a film cap attached to a high voltage secondary. It IS necessary to use a high quality cap with voltage rating sufficient to not break down in this function, also it's no quicker than a resistor to charge the supply. The reactance of the capacitor(s) is equivalent to a resistor wrt to charge rate. If you're looking for serious reliability use a 1KV or up rated film cap. I see them fail all too often in JCM900s.

There's another hangup with using a cap here, the supply will not charge the same if run on a different frequency. It will be insufficient if the amp is operated on a 50 Hz power supply. This will be much to the detriment of the amp's health if carried from USA/Canada to parts of the world that run on 50 Hz. I found out the hard way when wrangling a couple of early Sundowns on a recording session in London. Changing a resistor in the voltage divider is all that's needed to bring the bias back to where it should be.

One has to wonder, why did Fender do this? How many decades they've put a normal bias supply in their amps, not sensitive to line frequency, very reliable. I don't see there's any special charm in doing it this way, it certainly doesn't make for a unique tone or anything.

[Dave H]

Those two 0.1uf capacitors between the winding and the bias supply seem to be way too small for the application. Any chance those are actually 1.0uf?

You can raise the absolute value of the bias by increasing the value of those two capacitors. If they are indeed 1.0uf, replacing both of them with a single 1.0uf @ 630V capacitor will probably almost double the absolute value of the bias.

I think the 2 x 0.1u in series is correct. It looks like a voltage doubler with capacitor divider. Changing from 0.05u to 1u will pump up the bias to hundreds of volts (I think). I'd try adding a 0.1u in parallel with one of the existing 0.1u caps to start with. That should increase the bias voltage a little (to -50V?)

[66 Kicks]

The Ampeg bias is a different animal compared to the EC bias. The Ampeg bias uses a capacitor and a resistor to generate a really ugly AC signal which is then rectified and filtered. The EC bias generates a sine wave that is completely negative (except at zero volts) which is then rectified (doesn't seem necessary, does it?) and filtered. The magnitude of the sine wave depends on the value of those series input capacitors and the value of the capacitor connected between those input capacitors and ground. The filtering capacitor after the rectifier diode controls the ripple magnitude, but it is not a factor in the bias voltage. The load resistance is also not a factor in the bias voltage, within reason.

Both the Ampeg and Vox versions are for bridge rectifier power supplies with no bias winding. The EC version is for a conventional CT FW rectified power supply with no bias winding or tap. The Vox bias is much more similar to the EC bias in that it also generates an all-negative signal. The Vox version depends on the B+ supply for its voltage, so standby is a problem. The EC version doesn't depend on the B+ supply, so a B+ rail standby can be used. Besides, it will charge up lightning fast.

The impedance of a 1uF capacitor at 60Hz is 2650 ohms. Typical bias supplies have resistors in the tens of thousand to hundreds of thousand ohms in the charging path.

[tubeswell]

Those two 0.1uf capacitors between the winding and the bias supply seem to be way too small for the application. Any chance those are actually 1.0uf?

No chance. They are 0.1uF 600V 'suntan' caps in series (see pic) - giving 0.05uF@1200V

[tubeswell]

Other gutshots

[66 Kicks]

I now see where the discrepancy is. My notes have a 10uF capacitor in parallel with the diode to ground and you have 1uF. I also have 100uF for the filter capacitor and you have 10uF. My version will charge up much faster and it will rival the ripple of your version.

[g1]

Just for clarification, tubeswell is referring to the official Fender version: