Hiwatt model

[DLC86]

HI @FractalAudio, I think you're the only one who can help on this subject.
From what I've read early versions of Hiwatt Dr103 (the one I owned) had a cathode follower feeding the phase inverter. I assume the one modeled in the axe fx is one of the later versions with no Cf though, since cf comp and harmonics default to zero when loading that amp.
Which values should I set cf comp, cf hardness and harmonics to simulate the older design?
Here's the schematic of a late '60s Dr103 preamp:



Thanks in advance!

 

[DLC86]

A little update for those interested: yesterday I experimented a bit with Cf related parameters and found a combination that sounded good, then I had the idea to look at the schematics of amps that have a cathode follower and found out that vintage Marshalls like the jcm800 and jtm45 have a pretty similar design, so I looked at the models of those amps in the axe and found out that the default values are really close to what I had previously set by ear.
Basically cf comp around 70%, harmonics around 0.5, cf ratio at 0.450 and cf time 3.2ms.

Here's the schematic of the jcm800:
As you can see the circuit around V2 is similar to Hiwatt's V3, the only differences are the values of resistors that go from cathodes to ground (marshall 1k & 100k, Hiwatt 2,2k & 100k), the Hiwatt's presence circuit in the first triode of V3 and probably a different voltage reaching the plates= different bias point.
Sadly I don't know how these differences would affect the cathode follower behaviour or how to traslate them into parameters' values, probably only Cliff knows how to do it and I'm not sure he can even do it without actually taking measurements on the real amp.
But the values I wrote above definitely get in the ballpark

Disclaimer: I have limited knowledge of how a tube amp works so I could have just written bullshit nonetheless, I apologize in advance if that's the case

 

[AdmiralB]

There's no real reason to have a CF feeding the phase inverter. Typically they'd be used in amps where one needs a low-output-impedance current source to drive the next circuit. The PI is not a low-impedance target, so it's not an issue of loading as it is with a tone circuit, and there's really no reason to put a CF in front of it unless 1) you've got a triode section left over and can't think of anything else to do with it, and/or 2) you want the sound of a distorted CF.

The main difference between the Komet 60 and the Concorde is that the latter has a CF feeding the PI...in this case, it's for both of the above reasons. The 'stock' Express/K60 circuit has a leftover triode, and overdriving a CF adds a unique nuance to the tone, perhaps a bit more "Marshally". I built a Komet clone about ten years ago that could switch the CF in and out; when set clean, it really makes no difference at all.

 

[DLC86]

There's no real reason to have a CF feeding the phase inverter. Typically they'd be used in amps where one needs a low-output-impedance current source to drive the next circuit. The PI is not a low-impedance target, so it's not an issue of loading as it is with a tone circuit, and there's really no reason to put a CF in front of it unless 1) you've got a triode section left over and can't think of anything else to do with it, and/or 2) you want the sound of a distorted CF.

The main difference between the Komet 60 and the Concorde is that the latter has a CF feeding the PI...in this case, it's for both of the above reasons. The 'stock' Express/K60 circuit has a leftover triode, and overdriving a CF adds a unique nuance to the tone, perhaps a bit more "Marshally". I built a Komet clone about ten years ago that could switch the CF in and out; when set clean, it really makes no difference at all.

Well, probably it depends on the amp design but I think something's happening in clean amps too, or at least when they're at the edge of breakup. By tweaking the CF parameters in the axe I can clearly hear a difference even on clean sounds and even more with a dirt pedal set at unity gain in front of a clean amp, it makes the sound more "rounded".
Cliff himself stated in an old thread I read a few days ago that a cathode follower causes some strange interactions between tubes that go beyond impedance change and compression. I assume that what the harmonics parameter does has something to do with this

 

[AdmiralB]

When you tweak the parameters like that, you're doing something that isn't happening in the real amp.

In a Marshall - or in the Concorde design - you can't overdrive the CF without first overdriving the preamp. Kinda like setting the bias at 100% hot - you can do it in the Axe because it's not real.

 

[DLC86]

When you tweak the parameters like that, you're doing something that isn't happening in the real amp.

In a Marshall - or in the Concorde design - you can't overdrive the CF without first overdriving the preamp. Kinda like setting the bias at 100% hot - you can do it in the Axe because it's not real.

I have to disagree with this. Tweaking those parameters isn't equal to overdriving the CF.
In this article is explained very well how a cathode follower works http://www.valvewizard.co.uk/dccf.html

CF comp, CF time and CF ratio parameters basically set the asymmetric compression that the cathode follower introduces (I think you can see them as threshold, attack and ratio) and that compression is present even without overdriving the stage.
One of these parameters (presumably cf ratio) should be the same as reducing R4 value in the article's example.

Harmonics control is neither a gain control as it doesn't seem to introduce any clipping in the signal, it "distorts" it in the sense that it changes the ratio between the fundamental and the upper harmonics (probably also even/odd) but I don't think it's directly related to overdriving the CF, instead it is emulating some interactions occuring between the gain stages (CF included) and becoming more apparent as the gain is increased.
From the Fw Q2.0 release notes: "It controls the amount of interaction. Higher values yield softer distortion."

CF hardness is probably the only parameter directly related to cathode follower clipping but again I don't think it's a gain control, to my ear it works very similarly to preamp hardness, it basically varies the clip shape of that gain stage.

 

[AdmiralB]

So when you're screwing with the parameters...where, in the circuit, is the CF? You don't know.

In probably 99% of tube amp circuits that have CFs, they're either driving tone stacks or effects loop sends. You asked about a specific situation where it's driving a phase inverter. I gave you answers based upon experience with that specific situation. And you're telling me I'm wrong, because you're fiddling with parameters that control something that neither you nor I know anything about with respect to how it applies.

 

[DLC86]

So when you're screwing with the parameters...where, in the circuit, is the CF? You don't know.

This is a good point and that's why I was hoping for Cliff to chime in the discussion.

In probably 99% of tube amp circuits that have CFs, they're either driving tone stacks or effects loop sends. You asked about a specific situation where it's driving a phase inverter.

The compression and interactions introduced by the CF have nothing to do with the stage it's driving, if you'd read the article this would be clearer to you

I gave you answers based upon experience with that specific situation. And you're telling me I'm wrong

I'm not telling that what you've experienced with that amp is wrong but only that that particular design minimizes these effects and you were not hearing any difference on clean sound.

because you're fiddling with parameters that control something that neither you nor I know anything about with respect to how it applies.

I was just disagreeing on this statement: "you can't overdrive the CF without first overdriving the preamp". You assume that fiddling with those parameters is equivalent to overdriving the CF but that clearly is't correct, there's no need to be a genius to understand what those parameter do to the sound, it's pretty clear for example that CF comp introduces compression and not distorion, that CF ratio varies the ratio of that compression, that CF time varies the attack of that compression, no rocket science... And reading that artcle which explains how a CF works definitely help to understand what's the real world equivalent of those parameters

 

[DLC86]

Another update: after a few days I re-read that article and figured out that basically the Hiwatt CF doesn't introduce any compression, that's because the CF can't steal current from the previous triode's anode due to the resistor network between the two triodes (1M8 and 1M) and the larger cathode resistor (220K).
So the Hiwatt cathode follower is not DC coupled to the previous stage and it's just acting like a useless buffer.

Turned out that AdmiralB was right in this particular case but not for the reasons he was saying

 

[sonofmickel]

Have a closer look to try and spot the differences.
The Hiwatt is not too similar to a Marshall Cathode fed tone stack.
The Hiwatt is a plate fed tone stack with a cathode bypass cap.
The Hiwatt comes off of the 220k plate resistor, the Marshall feed comes off of the 100k Cathode resistor.
Look at Splawn, Splawn uses Plate Load tone stacks but have an extra gain stage(or two) up front and also a ton of treble snubber caps.

 

[DLC86]

Regarding the Hiwatt model maybe I've found an issue.
Yesterday I had the opportunity to do a comparison with a vintage Dr103 (I sold mine when I bought the axe fx) and that confirmed an impression I had since I started using our beloved modeler:
The modeled normal channel is way darker than on the real amp, it's just like it has a hi-cut applied.
On the real amp instead the only difference between the two channels is the bass response and that is somewhat confirmed by the schematic, what differentiates the channels are just a coupling cap and the cathode bypass cap on V1, these basically set the corner frequency of a highpass filter (if I understand it correctly).
To further confirm this it was super-easy to match the real amp's normal channel to the brilliant model on the axe, I just lowered Preamp Lo-cut to about 20 Hz and that's it!
Even the knobs' position matched almost exactly the amp while that didn't happen when selecting the normal model on the axe.
@FractalAudio?

EDIT: Probably I had done that comparison with the Jumped model, Normal model is correct

 

[DLC86]

Regarding the Hiwatt model maybe I've found an issue.
Yesterday I had the opportunity to do a comparison with a vintage Dr103 (I sold mine when I bought the axe fx) and that confirmed an impression I had since I started using our beloved modeler:
The modeled normal channel is way darker than on the real amp, it's just like it has a hi-cut applied.
On the real amp instead the only difference between the two channels is the bass response and that is somewhat confirmed by the schematic, what differentiates the channels are just a coupling cap and the cathode bypass cap on V1, these basically set the corner frequency of a highpass filter (if I understand it correctly).
To further confirm this it was super-easy to match the real amp's normal channel to the brilliant model on the axe, I just lowered Preamp Lo-cut to about 20 Hz and that's it!
Even the knobs' position matched almost exactly the amp while that didn't happen when selecting the normal model on the axe.
@FractalAudio?

EDIT: Probably I had done that comparison with the Jumped model, Normal model is correct

Finally found the culprit!
I've compared, using the tonematch block set up as spectrum analyzer, Hipower Normal model to the normal channel of Hipower Jumped model at default settings (except turning off treble drive in the jumped) and as you can see there's an evident high frequencies roll-off on the jumped model, while these 2 should be identical. I'm gonna open a bug thread, I'm not crazy!



EDIT: fixed in Fw Q6.02!

 

[Wolfenstein98k]

Great sleuthing!