OD trim voltage value

[talbany]

Not to come off as a butthead here...It looks to me like we are all kind of missing the point here..Analyzing how the plate resistor functions within the voltage divider network or Plotting transconductance curves and doing all the math is important in understanding how a circuit or componet functions...However the ultimate goal here is TONE!!.. How the tone and character changes w/ the differant values mentioned... For instance.. Several points were refrenced by a output as it relates to the value of the plate resistor.. It could be on paper that it does exist...Has anyone here noticed changes in volume, tone. feel with respect to output...To me this is much more intresting than going through a post with a bunch of Av = (mu*Ra)/(Ra+ra) = (100*100K)/(100K+62.5K)= 61.5
= 35.8dB..Just stirring the pot...


Tony VVT

[heisthl]

I used to have my TI electronics calculator on my bench so I could use it immediately, when I took it off my bench and quit using it altogether, my designs started to sound better.

[butwhatif]

Like Charlie Parker said, 'first master your instrument, then forget all that shit and play'. Or Miles-- ''i'll let you know what it is after i play it' -or something to that effect.

[dogears]

I kind of agree on some points with Tony and not others.

But, as I pointed out to him when we sparred regarding the knee frequency of a bypassed cathode stage, if you post info, make it correct.

To my ears, the high plates do have more output than low plates. Harder clipping too, but they drive the stages in a cool way and have assymetric clipping. I dig them. 100K plate amps definately have less output to me and have softer cliupping and more preamp OD.

[dogears]

Of course I have noticed huge changes due to the output differences. The gain staging is totally altered as you hit the next tube harder. As a result, you run a lower pre-OD trimmer.

Sorry to be a reverse butthead, but the discussion of output it pertinant. I for one do like to know why things respond the way they do. Kind of like putting a face to the name.

Not to come off as a butthead here...It looks to me like we are all kind of missing the point here..Analyzing how the plate resistor functions within the voltage divider network or Plotting transconductance curves and doing all the math is important in understanding how a circuit or componet functions...However the ultimate goal here is TONE!!.. How the tone and character changes w/ the differant values mentioned... For instance.. Several points were refrenced by a output as it relates to the value of the plate resistor.. It could be on paper that it does exist...Has anyone here noticed changes in volume, tone. feel with respect to output...To me this is much more intresting than going through a post with a bunch of Av = (mu*Ra)/(Ra+ra) = (100*100K)/(100K+62.5K)= 61.5
= 35.8dB..Just stirring the pot...


Tony VVT

[bluesfendermanblues]

I for one do like to know why things respond the way they do. Kind of like putting a face to the name.

[martin manning]

In the spirit of discovering why and putting a face to a name, I made some plots to show what happens to the load lines and operating points as the plate load is changed.

The first plot shows the anode load lines for 100, 150, 180, and 220K Ra on a representative set of grid curves. There are also two cathode load lines, one for 1K5 Rk and one for 3K3. The symbols show the quiescent points for the typical combinations of Ra and Rk, namely 100K-1K5, 150K-2K2, 180K-2K7, and 220K-3K3. The anode and cathode load lines will stay pretty much where they are for any combination of Ra and Rk, so you can see how the quiescent point will move on the characteristic for various non-standard pairings.

The second chart shows the resulting gain and %HD for the typical combinations of Ra and Rk.

First, there is an increase in gain for the higher plate loads, but it is only a dB or so. The %HD more than doubles, and it would seem that this is the more significant effect. However, note that the %HD calculation used here is the simple "Five Equal Ordinates" method from RDH4, and is done "at clipping," where the signal is assumed to take Vg-k to zero on the positive swing along the AC load line. This means that the higher plate cases are assuming a larger signal than the lower ones, and the negative swing is driving them closer to cut-off by two times the difference in the quiescent point Vg-k, which increases the calculated distortion.

So what if one were to bias all of the cases at the same quiescent Vg-k, say the -1.95V from the 220K-3K3 case, by adjusting the value of Rk? That's the third plot. Note Rk is now ~2K for the 100K Ra instead of 1K5. The gain effect with increasing Ra is about the same, but the calculated %HD is now essentially flat at the level of the 220K-3K3 case.

Conclusions:

The difference in gain going from 100 to 220K Ra seems relatively small, on the order of 1 dB for one stage. Cascaded stages would of course multiply this effect.

The shift in the bias point is significant to the distortion produced, and therefore so is the combination of Ra and Rk chosen. There are of course many other thing going on in a whole amp, but it may be that the movement of bias point is as significant or more so than the increase in gain in the perceived output.

MPM

[markusw]

Thanks a lot!
Interesting read....will take some time to completely understand though

[bluesfendermanblues]

Thanks to Martin Manning for posting a thorough analysis of various preamp resistor values. I cannot confirm the conclusion that preamp resistor values does not have a substantial difference in dB output.

I did a test on my 50w non-hrm amp, recording different resistor values into my cubase 5.

My finding was, that in the clean channel (wtihout OD) low plates (100k/1.5k) produced a 4dB higher output than high plates (220k/3.3k//150k/2.2k) - And the high plates had more low end that the low plates.

I used a SM57 up close, with master volume on 1½ and I didnt touch a dial. In OD mode 100k was (as a consequence of higher output) more distorted than 220k setup.

I don't claim to be an expert, but these were my findings. It could be interesting if someone else did this test.

[martin manning]

- And the high plates had more low end that the low plates.

If you don't change the value of Ck or the coupling cap (Co) to the following stage, and you double the values of Rk and Ra, then the low frequency roll-off at -3 dB for Rk-Ck and Zo-Co will both be halved. That might explain the "more low-end" part of your result. The response characteristics of the rest of the system, including your speaker and mic, are mixed into the result too.

I am in no way trying to say this is simple, just trying to better understand a little piece of it :^)

MPM

[talbany]

My finding was, that in the clean channel (wtihout OD) low plates (100k/1.5k) produced a 4dB higher output than high plates.

At what frequency?...Remember 1.5k cathode changes bias greatly altering frequency responce.. I hear brighter...A shift in frequency responce should not automatically be assumed as more output..Since our hearing is more sensative to mid-high frequencies and in most cases appear louder...To be honest with respect to output it would need to be charted over a wider frequency range...If you have one would prove your statement without a doubt.. See MM's conclusion's

There are of course many other thing going on in a whole amp, but it may be that the movement of bias point is as significant or more so than the increase in gain in the perceived output

--------------------------------------------------------------------------------

In the spirit of discovering why and putting a face to a name, I made some plots to show what happens to the load lines and operating points as the plate load is changed.

The first plot shows the anode load lines for 100, 150, 180, and 220K Ra on a representative set of grid curves. There are also two cathode load lines, one for 1K5 Rk and one for 3K3. The symbols show the quiescent points for the typical combinations of Ra and Rk, namely 100K-1K5, 150K-2K2, 180K-2K7, and 220K-3K3. The anode and cathode load lines will stay pretty much where they are for any combination of Ra and Rk, so you can see how the quiescent point will move on the characteristic for various non-standard pairings.

The second chart shows the resulting gain and %HD for the typical combinations of Ra and Rk.

First, there is an increase in gain for the higher plate loads, but it is only a dB or so. The %HD more than doubles, and it would seem that this is the more significant effect. However, note that the %HD calculation used here is the simple "Five Equal Ordinates" method from RDH4, and is done "at clipping," where the signal is assumed to take Vg-k to zero on the positive swing along the AC load line. This means that the higher plate cases are assuming a larger signal than the lower ones, and the negative swing is driving them closer to cut-off by two times the difference in the quiescent point Vg-k, which increases the calculated distortion.

So what if one were to bias all of the cases at the same quiescent Vg-k, say the -1.95V from the 220K-3K3 case, by adjusting the value of Rk? That's the third plot. Note Rk is now ~2K for the 100K Ra instead of 1K5. The gain effect with increasing Ra is about the same, but the calculated %HD is now essentially flat at the level of the 220K-3K3 case.

Conclusions:

The difference in gain going from 100 to 220K Ra seems relatively small, on the order of 1 dB for one stage. Cascaded stages would of course multiply this effect.

The shift in the bias point is significant to the distortion produced, and therefore so is the combination of Ra and Rk chosen. There are of course many other thing going on in a whole amp, but it may be that the movement of bias point is as significant or more so than the increase in gain in the perceived output.

..


Nice Post..



Tony VVT

[bluesfendermanblues]

My finding was, that in the clean channel (wtihout OD) low plates (100k/1.5k) produced a 4dB higher output than high plates.

At what frequency?...Remember 1.5k cathode changes bias greatly altering frequency responce.. I hear brighter...A shift in frequency responce should not automatically be assumed as more output..Since our hearing is more sensative to mid-high frequencies and in most cases appear louder...To be honest with respect to output it would need to be charted over a wider frequency range...If you have one would prove your statement without a doubt.. Tony VVT

the 4 dB was the reading on the cubase 'channel VU-meter' and pretty straight even over the whole frequency range.

[bluesfendermanblues]

- And the high plates had more low end that the low plates.

If you don't change the value of Ck or the coupling cap (Co) to the following stage, and you double the values of Rk and Ra, then the low frequency roll-off at -3 dB for Rk-Ck and Zo-Co will both be halved. That might explain the "more low-end" part of your result. The response characteristics of the rest of the system, including your speaker and mic, are mixed into the result too.

I am in no way trying to say this is simple, just trying to better understand a little piece of it :^)

MPM

Agree with ya....
- The high plate (220K, 3.3k + 150k, 2.2k) used 5uF
- the low plate used 10 uF

Hence, explaining the difference in low end (but not the overall volume difference)

[Structo]

Martin, thanks for the work on that.

But why did you choose 334v?

Isn't that like 34v over spec for a 12ax7?

Why not use a voltage typical for a Dumble which is around 185-210v?

And, was that real world testing or software modeling?

Thanks

[martin manning]

Agree with ya....
- The high plate (220K, 3.3k + 150k, 2.2k) used 5uF
- the low plate used 10 uF

Hence, explaining the difference in low end (but not the overall volume difference)

So you did (partly? mostly?) compensate for the resistance changes. The 220K stage break point is about the same, and the 150K stage is down by about 30%.

Structo, that's all theoretical, calculated with Excel.

Note that it's the B+ that is 334V, the anode voltages at the quiescent points are in the 200-220 range as you can see on the load line plot.

MPM