Distortion frequencies

[Normster]

I read a comment in another forum about using different plate resistor values to change where clipping occurs. Is that why Dumble's plate resistor values are all over the map? It makes sense that if clipping occurs on many different parts of the waveform, the overall effect will be smoother distortion. Am I way off base here? Has anyone done a waveform analysis of various gain stages?

[glasman]

I read a comment in another forum about using different plate resistor values to change where clipping occurs. Is that why Dumble's plate resistor values are all over the map? It makes sense that if clipping occurs on many different parts of the waveform, the overall effect will be smoother distortion. Am I way off base here? Has anyone done a waveform analysis of various gain stages?

Hi Norm, not sure if that is what Dr. D. was shooting for with the resistor values. I have scoped the circuits a lot over the last 3 years and have found for the most part there is minimal if any clipping in the first two stages. Stage 3 starts the clipping (on one side) and Stage four clips the other side of the wave form.

I would suspect that the Dr. might have varied the values to to control the gain and frequency response of the various stages. Subtle difference in plate values can have a large effect on the overall tone of the stage. The bigger plate resistors reduce the high frequency response of the amp.

I have not tried it, but would suspect that an amp with 100K/1.5K plate/cathode pairs through the entire amp would have more chime than say a 220K/150K setup. This would be a good experiment for someone to try (with the right measurement equipment).

Just my 2 cents.

Gary

[BobW]

A Spice circuit or other circuit simulation would only provide minimal results since the critical components (tubes, transformers, speakers) are difficult to model correctly.
Norm, Gary, I believe the only real way to accomplish this is to demonstrate A/B tests on a real amp using the ol ears and a spectrum analyzer to validate what is really being heard. If anyone has a spectrum analyzer for sale...cheap, I'm interested.

[dogears]

There is a corollary effect of the larger plates in that they have correspondingly sized cathode resistors. The cathode resistor has a direct effect on the bass response due to the bypass cap/cathode resistor relationship. A 220k stage will have much more bottom than a 100k stage, assuming same size bypass and RCA value cathode resistor sizing.

[BobW]

The cathode resistor has a direct effect on the bass response due to the bypass cap/cathode resistor relationship. A 220k stage will have much more bottom than a 100k stage, assuming same size bypass and RCA value cathode resistor sizing.

I could agree more. To optimize the tone from stage to stage, it would be best to start with the standard HAD design, as you've suggested, and adjust from there. It was suggested the analyzer could help validate the changes were improving the tone based on the builders preference and the desired frequency response.
Scott, You have the best set of ears on this board, hands down, about how a HAD amp should sound, but for my ol ears, the analyzer would be a welcome tool to my bench, and it could also help end the mystification of his amps. Am not convinced HADs ears were really that great, but believe his tools and experience helped him achieve the results.

[skyboltone]

PC software based audio spectrum analysers are abundant and cheap.

http://www.trueaudio.com/rta_abt1.htm

In fact the BARS display on Windows media player is a spectrum analyser of sorts.

Dan

[Luthierwnc]

I suspect he also had a more consultative practice. You make it for what you think someone wants and then they tell you what they like or don't like. His clientelle were the sort who would know what they wanted done and not shy (or cheap) about getting it. sh

[Luthierwnc]

In the Funk book he mentions clipping in a four gain stage amp. This isn't the same as we have but Gary's assessment of the last two stages is what I have seen too. Running a 1 khz sine at moderate settings the waveform looks much the same at the clean master. Past that point, it clips the top and then the bottom. My wave distortion looks more like the edge of a samauri sword, though. When pushed, the tops will crater like the top of a cursive "r". Here is the diagram:

[glasman]

If you are willing to build an interface box. Speakerworkshop is a true spectrum analyzer. And will measure the transfer function of circuit. Quite a bit more accurate than an RTA.

The best part, its free. Of course you are going to have to tweak around with the interface box to get it to work correctly, but it will give amplitude and phase information using a pink noise source.

Gary

www.speakerworkshop.com

[BobW]

Sky, I can work most test gear w/o problems but am not too computer savvy. Have you tried it yet, any opinions? thanks

[heisthl]

This might be a ticket to "Grail" status showing up on your bench scope (for those of you that think you are hearing impaired). Could someone out there with what they consider a "Grail" amp put a known input freq at recreatable front panel settings and display the OD2 waveform on a scope, adjusting the input freq until the clipping waveform and level on the +/- peaks exactly matches. Ex. say at 2K the pos/neg peaks are identical, as a builder you could insert 2K and tweak until peaks match and Voila! grail tone. Easier yet inject the signal post cl2 and pre od1. I'm new to the forum but not new to building/mods. This is the best forum I've ever spent 2 weeks lurking on - I'm caught up now - I've finally read all the posts. My first project for this forum will be a friends Music Man Rd-50 chassis to non-HRM. Adding 2 preamp tubes to that chassis will be tricky....

[BobW]

Gary, I imagine the interface would only need to be something as simple as a voltage divider w/non inductive resistors, short leads maybe coax, etc? This would help minimize reactive/phasing errors. I'll read more about it, since I may not want to use just a pink noise source, will let you know. Thanks for the info.

If you are willing to build an interface box. Speakerworkshop is a true spectrum analyzer. And will measure the transfer function of circuit. Quite a bit more accurate than an RTA.

Norm, sorry your initial post got sidetracked.

[Bob-I]

In the Funk book he mentions clipping in a four gain stage amp. This isn't the same as we have but Gary's assessment of the last two stages is what I have seen too. Running a 1 khz sine at moderate settings the waveform looks much the same at the clean master. Past that point, it clips the top and then the bottom. My wave distortion looks more like the edge of a samauri sword, though. When pushed, the tops will crater like the top of a cursive "r". Here is the diagram:

If you look at this diagram, the waveform is clipped with the first 3 stages. My guess is that because Dumble loses a tone of gain in the tone stack that the first 2 stages act like the first stage in this drawing.

Interesting, I always thought that the first OD stage clipped and the second compressed. My bad.

[skyboltone]

BobW, no, never used it but am following this avidly. I have used RF spectrum analyzers mostly trying to locate out of band harmonics and spurs. For analysis of the carrier or the audio content of an AM or SSB signal a scope will do.

Heisthl, David Hafler once observed that good amps always test well but bad amps don't always test poorly. Any waveform analysis will fall far short of your ears in determining the quality of a signal. Otherwise we'd all be using sand state amps instead of hollow state.

.02

Dan

[Normster]

Man, there's so much I don't know! Thanks for bouncing around the ideas and theories. I started thinking about this in my "pre-Dumble" days after hearing a friend's rig. He used two Arion Tubulators in series and got this great smooth OD. Very violin like. Both boxes were set to different levels of gain and the effect was superb.

The last ODS I built (the 5AR4/6V6 reverb) used 100k Rk/1.5k Ck throughout like #124. I'll have to get it back for awhile to see if it's chimier, smoother, raspier, or what. Maybe build another with Hybrid-A values to compare the two.