More on bias etc.

[skyboltone]

The amp I'm currently working on has a pair of 7189/EL-84M Russian equivalents. I've used the Express values and scheme with the exception of a 1K PI bias resistor and a 39K tail. The power section is cathode biased with a 130 ohm resistor.

With 305 VDC on the plates the 130 ohm resistor is making about 9 volts + on the cathodes. I had initially planned to run this amp with fixed bias but I'm using a bridge rectifier for the main supply. As I'm sure you're aware you can't create a positive and a negative supply from the same PT winding unless it's a full wave with center tap. I've got five open fuses to prove it.

Anyway, the 9 volts on the cathodes computes to about 69ma combination screen and plate current. Pretty low. The amp sounds a little congested.......like it's got a head cold. I cobbled together a negative supply using a wall wart and a variac today and got the grids down to about -12.2 volts today and the amp really opened up and acted much more like it should.

So here's the deal. I want y'all to confirm my notion that making a bias supply using the filament transformer (I'm using 12.6V filaments....7189s in series) is a bad idea. Right? I'm already running the filaments at +50 VDC above ground anyway because it cuts the noise and hum so well, but marrying the grids and the filaments that way seems to me like inviting trouble. So that leaves adding yet another transformer (tear up the wallwart?) or putting in a 12.5W 175ohm adjustable cathode bias resistor.

Anybody have a thought? Thanks in advance for the replies.

Dan

[rooster]

Is this a trick question? Yeah, drop the value of the cathode resistor to 90-100 ohms and heat those tubes up.

[Fischerman]

I had initially planned to run this amp with fixed bias but I'm using a bridge rectifier for the main supply. As I'm sure you're aware you can't create a positive and a negative supply from the same PT winding unless it's a full wave with center tap.

I must not be following you. Look at the Express schematic. They just tap off of one of the HV windings with a reversed diode and a dropping resistor to get the bias voltage. It's only half-wave but it works fine. Or just lower the Rk as suggested but it sounds like you want to go fixed-bias (I think Richie or Dana has posted that they switched a 'Pool to fixed-bias and liked it even more).

EDIT: I didn't see the 'bridge'. Has anybody seen the bridge? Where's that confounded bridge. Sorry 'bout that.

[skyboltone]

Well,...not really a trick question. But, definitely a trick answer. I don't think I communicated what I'm thinking very well. Evenings do that. Squint these numbers for a moment:

[IMG:800:507]http://i32.tinypic.com/21mes6x.jpg[/img]

So there are two ways to get class AB1 push-pull out of a 7189 or EL84 type at 300VDC on the plates. As a fixed bias you can put -14 volts or so on the grids and have 28 volts peak to peak signal for 4% distortion. Well, KF didn't design for 4%. The signal into the power section is decidedly higher than 28V PP. I think I need to put the scope on my build and put that out there so we know. In any case, trainwrecks are designed to push the power section into clipping. Notice that with -14 on the grids the 0 signal plate and screen currents are in the neighborhood of 15ma while the current jumps to around 120ma at full roar. In pure class B the numbers are similar. Incidentally, these numbers are in order with those one would expect in a class AB1/2 or class B.

Now look at the cathode biased numbers. With a 130 ohm cathode resistor, still supposedly in class AB1 our 0 signal plate and screen currents rise dramatically to around 80ma. Never mind that in my build I'm only getting 69ma (calculated with 9 volts dropped across 130 ohms). The last one of these I built with JJ EL-84s this voltage was 11.4V or so measured. The amp was less compressed and more "open" and singing too. Now, as you so eloquently put it, I can LOWER the cathode resistor value to bias the tubes into the 80ma range that the tube data suggests but what is that likely to do for us? If you take a look at my initial post, I tried a fixed bias number of -12.5V and the amp's character changed from a "amp with a head cold" to one that was open and much more "express" like. If I lower the cathode resistor I will also lower the grid to cathode electrical separation and LOOSE headroom. Not gain it. The result will likely be more compression and more tightness and nasal congestion.

What I'm suggesting instead is to RAISE the cathode resistor value, further lower the quiescent current, and gain the swinging room across the grid so that the tube isn't clipping so hard. Now, I realize that I didn't say that. I do that alot. I'm having a conversation with myself in my head and I figure you've been listening in all the while and know where I'm headed. Not the case is it?

Anyway, noodle that. I've got some adjustable cathode resistors coming and I'm going to tear apart that wallwart and see if I can squeeze it in on the power supply board somehow.....that is unless one of our brain trust knows how to get a negative supply out of power supply with a bridge rectifier. (hmmmmm dual output DC-DC converter powered from a voltage divider tapped off the screen supply......)

Dan

[skyboltone]

Double dang post. Allyn, let us delete posts.

[skyboltone]

I must not be following you. Look at the Express schematic. They just tap off of one of the HV windings with a reversed diode and a dropping resistor to get the bias voltage. It's only half-wave but it works fine. Or just lower the Rk as suggested but it sounds like you want to go fixed-bias (I think Richie or Dana has posted that they switched a 'Pool to fixed-bias and liked it even more).

EDIT: I didn't see the 'bridge'. Has anybody seen the bridge? Where's that confounded bridge. Sorry 'bout that.

I'm using a torroid power transformer with two identical sets of windings on either side of the core. 117/234 in 117/234 out. The only way to get my plate voltage with that guy is use a bridge rectifier. If I ground the center of the secondary I'll get 117-0-117. Not good. The PT also doesn't have a filament winding so I'm using a separate 12.6 transformer with the power tube filaments in series. Do you see the bridge yet? Now, if you put two bridge rectifiers in and ground the negative on one and the positive on the other they short each other. If you just put a diode in off one side of the AC you've got no reference to ground except through the main ps bridge. Also no good. I searched the literature on this and found no way to get a positive and negative supply off the same ungrounded winding. If you know a way please let me know.

[Wayne]

Hello all. Been lurking here for a while. Got lots more to say when I've got time, but I registered an account to respond to this post.

No problem to make a separate bias supply when the HT uses a bridge. Look at the way it's done in the Marshall JCM 900 (and, I'm sure, many others - I just happen to know that one 'cuz I'm cursed with a 900)

http://www.schematicheaven.com/marshall ... w_4100.pdf

Check out C15 (the capacitor marked "class X" on the power board) and everything to it's left. "Class X", IIRC, means a 1000 V component.

Looking forward to sharing/learning with you all.

W

[skyboltone]

Hello all. Been lurking here for a while. Got lots more to say when I've got time, but I registered an account to respond to this post.

No problem to make a separate bias supply when the HT uses a bridge. Look at the way it's done in the Marshall JCM 900 (and, I'm sure, many others - I just happen to know that one 'cuz I'm cursed with a 900)

http://www.schematicheaven.com/marshall ... w_4100.pdf

Check out C15 (the capacitor marked "class X" on the power board) and everything to it's left. "Class X", IIRC, means a 1000 V component.

Looking forward to sharing/learning with you all.

W

By gosh you're right. That's what I was looking for. I recall seeing a bias scheme a million years ago called a capacitor/resistor divider network that used a cap to isolate from the main supply but I think it was on the DC side. In this Marshall scheme they use a DC blocking .047 cap to isolate the diode, then filter and arrange the bias adjusting scheme in the regular way. I tried something similar with an .047 on the grounded end yesterday. Doh! It blowed up the 100uf filter cap. Thank you sir! And welcome to this crazy outfit.! I'll throw one of these together tomorrow. I think the 1000V rating is more about the AC voltage on the transformer. With 234 AC I imagine a 600volt cap would work.

Dan

[jjman]

For the cathode resistor, if you go that route, you can use a 5watt ~100ohm resistor in line with a 100ohm wire wound pot. This way the pot will dissipate less wattage and it's easy and cheap to find. Also you won't be able to accidentally short the cathodes to ground with a 0 ohms setting. I'm using that type of setup, but with one pot for each side, on my 6v6s on my "Express."

[skyboltone]

For the cathode resistor, if you go that route, you can use a 5watt ~100ohm resistor in line with a 100ohm wire wound pot. This way the pot will dissipate less wattage and it's easy and cheap to find. Also you won't be able to accidentally short the cathodes to ground with a 0 ohms setting. I'm using that type of setup, but with one pot for each side, on my 6v6s on my "Express."

Good idea JJ. I'll keep you posted here on the progress. I will say though that fixed bias with the 7189s is really a great way to go with these amps. When I get the bias situation squared away I'll probably put some trim pots in place of the PI bias and tail resistors and fine tune the sweet spot.

Dan

[Andy Le Blanc]

howdy... I was curious if you bypassed the 130 ohm bias resistor
the numbers you gave at the outset seemed to make 10.5 watts static dissipation
per tube which seems ok for a tube with a design-center of 12 watts.
I recently had an amp that I was reluctant to bypass for the tone
but it was nessesary for gig performance..... same issue..... "head cold"

[skyboltone]

howdy... I was curious if you bypassed the 130 ohm bias resistor
the numbers you gave at the outset seemed to make 10.5 watts static dissipation
per tube which seems ok for a tube with a design-center of 12 watts.
I recently had an amp that I was reluctant to bypass for the tone
but it was nessesary for gig performance..... same issue..... "head cold"

Hi Andy. It has a 250uf at 50V bypass cap. That's a KF value. Big. Also, how does one explain the 3 watt static dissipation for class A1 when using fixed bias? Misprint? It's in all the tube manufacturers tables. ?!! What I'm puzzled about is the apparent discrepancy between grid v.s. cathode voltage depending on bias method. I can tell you without qualification that the fixed bias at 12.5+/- sounds WAY better than cathode bias at 130ohms given the same PI values shown on the Liverpool scheme. At least with these tubes.

Also, class x (when applied to capacitors) does not mean 1000V component unfortunatly. On page 2 of the Power Amp CCT the same bias circuit is shown again, magnified, and the cap is called out 47n/250V with the "positive" end of the cap marked with an "R". <edit> class X is a type of noise suppression capacitor for "across the line" use where failure will not result in electrical shock hazard.<end edit>

Also note:! It's no mistake that the input values to this bias scheme are 56K and .047uf. Calculate the corner frequency of that network!!

[Andy Le Blanc]

I did some digging as to why there would be an apparent discrepancy
between grid v.s. cathode voltage depending on bias method
the 7189 may be a sub for a 6BQ5\el84 but it is not identical
It has a mu-factor rating between grid No.2 and grid No.1
this would be by design and would affect the method of bias as you've
demonstrated

[Andy Le Blanc]

Ive found reffence to the 7189 being optimized for what seems ultra-linear
hook up in RC-25.... RC-21...... RC-20 .....chathode biased

[Wayne]

Also note:! It's no mistake that the input values to this bias scheme are 56K and .047uf. Calculate the corner frequency of that network!!

Never noticed that before. I wonder if that's a coincidence? I can't think of an obvious advantage to having the corner freq right on like that. Seems to me better filtering would be had by setting it lower.

I guess, though, that with the amount of current flowing it doesn't matter.

I think to much

W