PSU observations and questions

[Steve Conner]

108v is pretty low, except if you're measuring at the actual plate. The node voltage you ought to be measuring is at the filter cap end of the plate resistor.

Of course your ears are the final judge, if you really have 110v at that supply node and you like the tone then it's all good! You might be getting some of the same mojo that EVH got with his variac.

[daz]

I measured at the plate. The node gives 180.

Thinking out loud here, i wonder if i'd be better off removing the 27k from the PI node since i want the cleanest possible PI. The first stage has a 10k, but i'm thinking that should go at the PI and the 27k's at the other nodes. I have anode for each of the 3 Ax7's. what'chall think?

[martin manning]

So, plot that load line on the graph I posted. I don't know what your Ra and Rk are, but for a 12AX7 and using the 100K and 1.5K values it goes from 180V on the x-axis to 180V/101.5K=1.77 mA on the y-axis (including the cathode resistor this time to be accurate). It crosses the cathode load line at the left end, right at the intersection of the 1.0 Ec line, and the plate voltage is ~115. Pretty close to what you are seeing. You now have only ~1V on the positive side to saturation.

What are the voltages down the rail now at each tube? You can plot the load lines for the gain stages knowing only the supply node voltages and the resistor values. Then measure the voltage at each plate relative to the cathode and use that to locate the operating point on each load line.

Plotting a PI load line is a little harder. You have to measure the plate voltage relative to the cathode, and calculate the plate current using the voltage drop across the plate resistor and its value. That plate voltage and plate current locate the operating point on the plate characteristic chart, and connecting that point to the supply node voltage at zero current is the load line.

Now you can see where your PI and the other stages are actually operating with respect to the grid voltage curves and decide if you want to change anything.

MPM

[Steve Conner]

If you're only dropping 80V across your plate resistor, the tube is running rather cold. What values are you using for plate and cathode resistors? Usually if you have a 100k plate resistor you'd use a 1.5k cathode resistor, and 2.2k (I think?) for a 220k plate resistor.

[daz]

If you're only dropping 80V across your plate resistor, the tube is running rather cold. What values are you using for plate and cathode resistors? Usually if you have a 100k plate resistor you'd use a 1.5k cathode resistor, and 2.2k (I think?) for a 220k plate resistor.

V1a: 100k plate, 820 ohm cathode. Tho i have tried up to 3.3k on the cathode. It just sounds richerer with the 820.

V1b: 100k plate, 1.5k cathode, tho also up to 3.3k. Again, seems to sound best like this, tho difference is minimal

No bypass cap on V1b, .68uf on V1a. Thats been as high as 10uf. loses too much clarity. 1uf is ok, tho for some reason driven sound is fuller with the .68 than anything bigger ! (talking bigger BEFORE the size that causes loss of clarity too bad)

So, plot that load line on the graph I posted

Not sure how

[martin manning]

I had already put a 180V supply/100K Ra/1.5K Rk load line on the plot.

What are the supply node voltages for your V1A and V1B? I don't know which one the 180V goes with.

MPM

[CitizenCain]

Not sure how

Here's a great little primer on messing with load lines. Broke the darkness for me.

http://diyparadise.com/tubeloadline/tubeloadlines.html

[daz]

What are the supply node voltages for your V1A and V1B? I don't know which one the 180V goes with.

things have changed a bit as i tweaked a few things. I put the 10k node R at the PI now and took that 27k and put it at V1. So things are now as follows....


V1 node:203v node R is 27k
Plates both 100k. v1a 110v, v1b 130v
v1a cathode 820 ohm/.68uf. v1b 1.6k, no bypass cap

v2 node:248v node R is 27k
plate 100k, v2b no plate R (cathode follower)
v2a 134v
820 ohm cathode on v2a and 100k on CF

PI node: 358, 10k R
plates are at about 220v
typical marshall PI values

Here's a great little primer on messing with load lines. Broke the darkness for me.

http://diyparadise.com/tubeloadline/tubeloadlines.html

Thats sounds great, but i'm in a rush to go to lunch and will read it when i get back. Thanks to both of ya...

[martin manning]

Still need the cathode voltages for the CF and PI. If you've got it open, can you fill this table in?

Each row is: Supply node V, Plate V, Cathode V, Plate resistor value, Cathode resistor value

Type Loc VB+ Va (pin 1 & 6) Vk (pin 3 & 8) Ra(K) Rk(K)
Gain V1a 203 110 2.5 100 0.82

Gain V1b 203 130 2.5 100 1.6

Gain V2a 248 134 1.2 100 0.82

CF V2b 248 248 ?? 0 100

PI V3a 358 220 ?? 82 (Rk not needed)

PI V3b 358 220 ?? 100 (Rk not needed)

The numbers you've given are filled in (?? are missing numbers), but it is best if you get all the measurements at the same time. I put together an Excel file that calculates the missing pieces and plots all the load lines. Might prove to be be a useful tool...

MPM

[daz]

I can't now because it's at work and i won't be there till monday. If i'm out that way this weekend i'll stop in and do it. otherwise i'll post the numbers next week. But i'm not quite sure what you want other than the cathode voltages. The voltage on V2b is the same as the node because theres no plate R. Like on a marshall with a CF....the plate of V2B goes straight to B+.

[tubeswell]

I put together an Excel file that calculates the missing pieces and plots all the load lines. Might prove to be be a useful tool...

MPM

Hi Martin

Is this a generic type spreadsheet or just one for Daz's current project?

If it is generic I wouldn't mind getting a hold of a copy, if that's okay with you.

[martin manning]

Daz, yes the only thing missing is the cathode voltages, and they are only really needed for the CF and PI, but if you get all of them then that allows some cross-checking and comparison to the expected values on the tube characteristic. My suggestion is to take all the readings again, at the same time, because they will drift some and it would be best to have a consistent set. No hurry; whenever you get the time.

The idea for the Excel sheet is that with a few measurements one can see where each tube is operating and what the load line looks like. It doesn't design the stages, rather it shows what a working circuit is doing. As an example, the attached plot was made from a Marshall 2204 schematic with voltages. Each load line has three symbols, the middle one being the quiescent operating point.

The plot shows that the first two gain stages are biased toward cut-off, especially Gain 2, with its 10K cathode resistor. Gain 3 and the CF are over near saturation, and the PI is more center-biased. You can tell pretty quickly where a gain stage is from the measured cathode voltage, but with CF's and PI's its not so obvious, and an automated tool makes it all very quick and mistake-free.

Tubeswell, I'd be glad to share, but I'd like to be confident that it is working correctly, clean it up a bit, trap input errors, etc. Let me fool around with it for a few days first.

MPM

[daz]

Ok, i got em....

V1a: .749v
V1b: 1.181v
V2a: .913v
V2b: 134.8v
V3: 33.20v

[martin manning]

Okay, here's the picture. Looks reasonable to me... if it sounds good, I'd leave it where it is. A bigger cathode resistor on the CF would move its operating point to the right, but I don't think a CF needs much headroom.

What do other folks think? Is this view of it informative?

MPM

[daz]

Thanks. I have to be honest at the risk of showing how clueless i am, but i really don't understand what i'm looking at or what i am supposed to deduce from it. All i know is that with you or anyone who DOES understand load line saying it looks fine, it reassures me that i not running outside of what would be considered reasonable parameters.