adjusting bias in 5G9

[tubeswell]

Sorry for the length of this post

A (rather lengthy) design question before I start my build - If I want to make the bias adjustable, (because I've had to get a custom-made PT and I won't know, until I hook everything up and get it going, exactly what I'm going to end up with in terms of B+, despite having asked for it to be wound to put out 300-0-300), then can somebody please confirm for me whether it is the 82k resistor I replace with a (100kN?) pot (with only wiper and input wired) in order to adjust the bias on a 5G9?

I've seen different explanations of an adjustable bias circuit on the Hoffman site here:

http://www.el34world.com/charts/Biascircuits.htm

But those circuits ain't quite the same as the 5G9 circuit. There doesn't appear to be a "bias range resistor' in the 5G9, (or is the 82k the equivalent of this?) Hoffman's bias circuit only has one bias capacitor, the 5G9 has two - one on either side of the 82K - is that because it needs to be decoupled from the 250k Depth control? (I gather if I use 10uF 450V caps in here instead of 8uF 150V caps, it won't matter?)

Also if I read it right, I'm trying to get -27.5V at each 6V6 Grid (and not -22.5V?. At least I think it says -27.5V?) The lettering on this .pdf version of the schematic I got from the Fender Amp Field Guide site is a little fuzzy.

http://www.ampwares.com/ffg/schem/tr..._5g9_schem.gif

Furthermore, can somebody explain to me the relationship between current, and voltage swing and resistance for signal at the output tube grid? and kindly edjamacate me about how I calculate the appropriate signal strength to drive the 6V6 s properly with this amp, and what setting the grid bias has got to do with this?

I have learned from when I built my 5F2A that I have to bias the output tube correctly to get the appropriate dissipation - But that amp is cathode biased. It runs a little hot but it seems to work fine. (On that amp there is 19V on the cathode and 344V between the cathode and the plate and with a 470R cathode resistor giving 13.76W. The plate is seeing about 40mA) I understand I could bring that down to a more desirable 12.6W if I changed (lowered? or raised?) the cathode resistor, or dropped the B+ about 30V.

What I want to understand is how this differs from a PP amp with fixed bias?

I know its got something to do with that you have two cathodes in parallel at ground potential in the 5G9, and that each grid needs to be at a negative voltage with respect to each cathode.

Also I understand from plodding through Jack Darr's book, that the gain from the previous stages results in the signal voltage going into the Output tube grid, and that this signal voltage needs to be of a given quantum that is somehow related to the plate and screen voltages to make the tube work optimally, but I'm not sure how to calculate this voltage, nor what the relationship of current and resistance is in this?

I have read that putting grid stoppers here will prevent 'too much current' draining from the PI into the output tube grid, and I can intuitively comprehend why you wouldn't want to heat up the grid too much (because then the grid starts acting like a cathode - right?). But is there another reason why don't you want to draw too much current from the PI (oh wait - is that because you don't want to overheat the PI plates? or is there something else?)

And is there an easier way to calclate current flowing through pin 8 in a fixed bias amp than putting a fancy gizmo between the tube and the socket that measures exactly what the current is? (like simply measuring plate and cathode voltages and using the resistance of the cathode resistor to determine what the current draw and dissipation of the tube was in the SE amp? - albeit that there isn't a cathode resistor in a fixed bias PP amp?) Is that what those lugs are that I've seen on some the back of some amps? - contact points for either side of pin 8 to plug your ampmeter into?

[MWJB]

On the 5G9 schematic replace the grounded 56K resistor in the bias circuit (after the 82K resistor) with a 50KL trim pot wired as a variable resistor, in series with say a 39K to ground.

10uf/150v caps in the bias circuit will be fine.

Don't think about an exact minus voltage dc at the grid, it's jst a means to an end, you want a voltage that gives satisfactory plate current on the 6V6s, usually 25-30mA at the voltages on the 5G9 schem.

12.6W on your 5F2A is only preferable if it sounds better, anything from 11W up to 20W is a feasible plate dissipation for an SE 6V6 amp like this. 14-16W are perhaps more typical. Fender never changed the value of the 470ohm cathode resistor as voltages were increased, nothing broke so they never revised it.

No diffference between PP and SE really, the real difference is between fixed & cathode bias. Fixed bias is typically set to a lower idle current to allow for the current to rise under load (gives better dynamics). Cathode bias typically runs more current but won't allow current to rise appreciably under load, causing plate voltage to drop, accentuating the envelope of sag & decay.

Build the amp stock, then report back as & if you have any issues, don't start trying to fix it before it's even broke.

Install a 1ohm (measured at exactly 1ohm) 2W resistor between each 6V6 cathode and ground. You can then read plate current in mV (mV converts to mA with this method) at pin 8 of each 6V6 socket. Measure this with the tremolo off (current will fly about with tremolo engaged). Otherwise, normally the 6V6 cathode is grounded on a fixed bias amp and there is no voltage to read.

no such thing as a "dumbass question", it's how we all learn...it's the answers you need to watch! :-)

[tubeswell]

Thanks MWJB

One more Qn then!

The PT I had built for me has a separate 50V winding for bias supply. If one end of the 50V winding is running straight into the 1N4007, will I get -69VDC on the other side?, and how is that calculated? (Sorry that's 2 more questions)

[tubeswell]

For MWJB

Shucks yet another Question!

Do I take that mV reading from pin 8 to ground? (I take it?)

Cheers

[MWJB]

"The PT I had built for me has a separate 50V winding for bias supply. If one end of the 50V winding is running straight into the 1N4007, will I get -69VDC on the other side?, and how is that calculated? (Sorry that's 2 more questions)" Possibly, but you will have a voltage divider downstream of the 1N4007 that will set the negative voltage to your desired level. To calculate the effect of the voltage divider you first of all divide the value of the load resistor by the sum of the load resistor and the dropping resistor. Multiply the answer by your anticipated -DC voltage at the 1N4007.

So let's say you have 370vdc at the 6V6 plates and want -28vdc at the grids (as per the schem).

You have a 56K load resistor grounded after the second 10uf/8uf filter cap, fed by an 82K dropping resistor. And we'll assume that you really do have -69vdc after the 1N4007

56/(56+82K)*-69vdc = V out

56/138* -69vdc = V out

0.4*-69vdc= -28vdc

If your B+ voltage goes higher, you will need more negative voltage (smaller dropping resistor) to prevent the tubes from burning up.

Yes, you take the dc mV reading at the 1ohm 6V6 cathodes to ground.

[tubeswell]

Follow-up for MWJB

Hi MWJB

Having just finished the amp I made it with adjustable bias as per your suggestion earlier. Before I fired it up I set the bias load resistor setup (i.e. 50k variable resistor in series with 39k) at 56k. The measured grid bias I get with this is -23V (as opposed to -27V on the schematic). The plate voltages are about 371 each, screens are about 1 volt lower. The cathode current measured over the 1R cathode resistors is about 31mA at idle. The amp sounds good and will go very loud, and has good distortion at full throttle, and is very clean at lower volumes. Should I aim to get the grid bias to -27V, or do you think it'll be okay to leave it at -23V? (or should I be aiming to get it above -25V? on the basis that I read somewhere in Jack Darr's book that the voltage swing on a typical 6V6 grid is about 25V - Did I understand that right?, or would he have been meaning peak-to-peak voltage swing?, which I think would mean that the voltage swing would never be enough to drive the grid into the positive? or doesn't it matter too much here?)

Is the calculation for dissipation for a fixed bias amp the same deal as for cathode bias? (i.e.; the cathode-to-plate voltage divided by the tube current, i.e. 371 x .031 = 11.5W per tube? This would seem slightly on the tame side.)

[MWJB]

Hi Tubeswell,

The plate current is the more important criteria, the negative voltage at pin 5 is just a means to an end...could be anywhere from -20 to -50v depending on the plate voltage that the 6V6s are running at. It's the plate dissipation that's goint to kill your tubes if too high, or make them sound thin & scratchy if too low.

Your tubes are running at 85% dissipation, typically a fixed bias amp will normally be set to anywhere from 50-70% (if they still sounded thin at 70% then you may need to increase plate current) BUT if you have good quality 6V6s (JJ, RCA/other robust NOS types) then they will be happy enough exactly as they are now. In fact, that's just how I would run them.

If you have the Russian Saratov 6V6s (Tung Sol/Sovtek/EH/GT6V6-R), I have little experience with these, but have heard recomendations that they last better with more conservative currents.

[tubeswell]

Thanks MWJB

They are the New Sensor 'Tung-Sol' 6V6GTs, which are supposedly made to take higher currents and voltages etc. But no one seems to understand much about this particular brand of 6V6 (maybe because they are 'newbies'). The one I have in my 5F2A runs fine at 13.7W (but that is cathode biased and has the same current at whatever load, or so I understand it).

So I rebiased them down to 27mA in my 5G9 (or so I think my meter read - it was swinging a bit in that ballpark, even with the trem controls cut) which took the plates down to about 363-4V (about 9.8W). I liked the sound of them better at 31mA 371V, because the amp actually had distortion when cranked. They weren't red-plating or anything (but would that just be a matter of time?). But now sadly, they sound a bit lamer and no distortion at full throttle .

Does higher negative voltage at the grids reduce/interfere with the voltage swing on the grids at all? They actually sounded really good before (DANG!). (Or is my meter simply not up to the task, and telling me lies, and were they all right before? Hmmmm....)

Or maybe I could go a bit higher with the tube current and get the plates nearer to 370V (whilst trying to keep dissipation around 10W)? Hmmmm... how much can I push this envelope?

[MWJB]

Bias should be set with the tremolo switched off.

"So I rebiased them down to 27mA in my 5G9 (or so I think my meter read - it was swinging a bit in that ballpark, even with the trem controls cut) which took the plates down to about 363-4V (about 9.8W)." - That's weird Plate voltage should RISE, not fall, when the tubes draw less current.

If you prefer the tubes at 30mA, then go with your ears. Even if they did wear out a little quicker, so what, they don't cost much.

Ignore the negative voltage at the grids, focus on the plate current.

[tubeswell]

Plate voltage should RISE, not fall, when the tubes draw less current.

If you prefer the tubes at 30mA, then go with your ears. Even if they did wear out a little quicker, so what, they don't cost much.

Ignore the negative voltage at the grids, focus on the plate current.

So plate voltage should rise aye? Hmmm... I did change the battery in my meter before rebiasing the amp, so maybe my I was getting a bad reading before?

I had a thought that the other possibility is that I haven't got a perfect 1R at each cathode. The meter (which could be at fault) showed these to measure around 1.6R and 1.7R (even tho' they are metal film resistors with 1% written on the packet), which is why I used 2 x 2R in parallel for each cathode.

Maybe if I used 10R at each cathode (or better still 2 x 20R, or 4 x 40R in parallel), I would get a more accurate reading? How high can I go with the cathode resistance in this way before the tube starts 'noticing' it is there? And what would the effect be? I suppose it is possible to run a part cathode-biased and part fixed-bias tube (albeit that the fixed bias would be lower than usual I guess)?

[MWJB]

10R would work fine, but so would some accurate 1R. With 10R at the cathodes you need to divide cathode voltage by 10 to get mA. Put the meter probes together and see what the display shows with zero resistance, maybe something like 0.4ohms?

Then when measuring the 1Rs subtract that 0.4R (for instance) default reading to help you select resistors that are 1R.

if you really have 1.6R at the cathodes, your mA readings will be artificially high.

[tubeswell]

Update on biasing

Thanks MWJB

I checked the meter and sure enough it showed 0.6 when the leads were crossed. I measured the cathode resistors again and sure enough they are 1R pretty smack on.

So I rigged up a plug for the trem footwsitch to knock out the trem, and switched it on. I got a steady reading of 28mA, so I took that back to 25.7mA and the plate voltage settled on 368, screen on 363.

When I took the current back to 20mA on one tube (we'll call it #1 tube) and the plate went to 375 and the screen went to 373-374, and on #2 tube the current was 19.1 and the plate was 373-372 and the screen was 374V!

So I took the plate current of the #2 tube to 22mA and the plate settled on about 371-372, and the screen settled on 370-371,so I'm happy to leave it with that.

So I'd conclude from that that even with the trem controls dialed way back, that it still needs the footswitch on to knock the trem completely out.

Anyhow now I will now do the ear test and report back.