Dual Pultec Frequency Response Curves

[zeeso]

Hello,
I finally finishd my Dual Pultec and thought I should share pictures of my finished unit with all of you. (sorry about the grainy pics, my camera sucks)

It works great and it sounds awesome, I was pleasantly surprised to see how well this can subtley shape the top end of a sound and the Lo end section really adds some balls to the music track. The unit itself is extremely quiet with no hum or noise in the output. I am very happy with the results and couldnt have asked for more from this project.

That said I have two questions

1 ) As you can see from the picture, the frequency response of the unit (all cut/boost controls at zero) is very flat except for a 1.5 - 2 dB bump around 25 Hz.
I think it could be the tube gain stage that is causing this bump as changing the frequency of the Lo boost switch from 20 to 120 Hz (with the boost set to zero) doesnt change the frequency of this bump even by a little. Anyone else seem similar behaviour on their Pultecs ?

2) The HT Voltage to the tube is 250 V exactly but the voltage on the tube side of the 4.7 uf Output Cap shows about 92 volts instead of the expected 125 V.
Does this mean that there could be something wrong with the Tube gain stage or tubes itself ?

By the way I have posted the frequency response of all the filters here if anyone is interested

http://s167.photobucket.com/albums/u123/zeeso/Dual%20Pultec%20Freq%20Resp/

Thanks for everyone's help and advise, this forum Rocks

Zeeso

 

[drpat]

...

 

[nerd]

Can't help you with your question, but...

Cool. That thing looks sweet.

Is that two gustav boards in a purusha case? I was going to attempt that, but decided to tackle some other stuff first. I think that may be next for me. I looks a little tight in that box. Anything I should be concerned about?

 

[hejsan]

Man, those curves look sweet.
what did you use to measure? some PC program?

I have to get one of purushas cases and get this over with.
Did you use 2 of gustavs boards?

thanks,
hejsan

 

[SSLtech]

What's your source and load impedances? -Might tell you something about the low end behaviour...

Keith

 

[zeeso]

I have my bypass switch wired so it bypasses the eq and amp section (the equivalent to connecting your xlr input cables directly to the xlr output cables) so it could be either the EQ section (which I am doubting) or the tube amp section or output transformers) I have used Fastron Inductors

Yes they are Gustav boards, I had to cut the EQ PCB tracks and modify them quite a bit to work with purusha's case. I does look a little tight but like I mentioned earlier I have had absolutely no hum or noise problems whatsover. I went with modding the EQ boards becasue the WIMA caps were PC mount and their leads were not long enough to go Point to Point. I think the unit looks much more clean and professional from the inside this way.

I used FuzzMeasure Pro (OSX App) to measure the frequency response curves.

I am not using an input transformer so my input is unbalanced but my output is balanced because of the output transfomer. Could this be causing the impedance mismatch ?
I measured this frequency response with a Protools 888 Interface balanced +4dBu Input and Output hooked up to this EQ.

Another thing to mention is that the 10K resistor from the input side of the 220n coupling cap to ground is not included on the PCB Layout so I did not include it in my pultec, should I wire it externally ?

Thanks

zeeso

 

[Purusha]

Hey Zeeso, nice work indeed!

Nice to see some tubes in those cases. Any chance to see some closer photos of the inside?
I am also considering of making a tube version this year but might try some other gain stage than
G-version... not that there is anything wrong with it.

BTW, I brought my solid state Dual Neveish Pultec into one big studio yesterday for the sound test
and the owner was very much surprised when putting it next to MP. Having the tube and the solid state
Dual Pultec might just bring you in the eq heaven. :grin:

 

[atticmike]

BTW: I fixed the low end bump for those who are interested.

 

[warpie]

Do you care to tell us how?

 

[Kingston]

I don't know why no one said it in this thread before but that low end bump is expected behaviour of the output cap working into the output transformer inductance. There's no fixing it without trade offs.

I personally deal with it by using a big enough cap so the bump moves below maybe 10-20hz, and perhaps compensating this in some earlier (decoupling) stage if possible. But really, who cares about a 2dB deviation at those frequencies.

 

[emrr]

Yeah.  I guess no one ever looks at all the tube gear measurements I have posted all over this place.  Even 20 years ago we were all supposed to be happy with transformerless gear quoted at +/-2dB 20-20K. 

 

[atticmike]

I don't know why no one said it in this thread before but that low end bump is expected behaviour of the output cap working into the output transformer inductance. There's no fixing it without trade offs.

I personally deal with it by using a big enough cap so the bump moves below maybe 10-20hz, and perhaps compensating this in some earlier (decoupling) stage if possible. But really, who cares about a 2dB deviation at those frequencies.

Why would there be a trade off? You just increase the cap, that's what I did and I'm at 16 uF so far, prolly 18 to be flat.

 

[Kingston]

Why would there be a trade off? You just increase the cap, that's what I did and I'm at 16 uF so far, prolly 18 to be flat.

The bump is not going to disappear is what I mean by a trade off, you can only move it around. And if I remember correctly from my own testing it will start behaving weird with very large values, depending on your transformers primary inductance. You can also somewhat affect it with a series resistor. Try 100 ohms (with a 500 ohm trimmer) and see what happens in both directions. Again, depends on the transformer what exactly will happen. Also, the bump is load dependent. Draw that graph with a 600-ohm load, then do it again with a huge cap and expect a big surprise.

If you have an amp with more than one decoupled stage, you can tune one of the decoupling RC filters so that its -3dB point is exactly at the center of the bump. That way response will stay flat a bit further, but low end roll-off will be more abrupt (now a 2nd-order filter). None of this actually matters in audio use at frequencies below maybe 20hz.

Also, this has been discussed extensively in many threads with the data posted, there's even an approximation formula to calculate the bump. I'll post the links here when I find those references.

 

[emrr]

Yes, I expect the shape to change with varying load. Fix it for one condition, blow up another condition.  This sort of gear is organic in behavior.

 

[atticmike]

Why would there be a trade off? You just increase the cap, that's what I did and I'm at 16 uF so far, prolly 18 to be flat.

The bump is not going to disappear is what I mean by a trade off, you can only move it around. And if I remember correctly from my own testing it will start behaving weird with very large values, depending on your transformers primary inductance. You can also somewhat affect it with a series resistor. Try 100 ohms (with a 500 ohm trimmer) and see what happens in both directions. Again, depends on the transformer what exactly will happen. Also, the bump is load dependent. Draw that graph with a 600-ohm load, then do it again with a huge cap and expect a big surprise.

If you have an amp with more than one decoupled stage, you can tune one of the decoupling RC filters so that its -3dB point is exactly at the center of the bump. That way response will stay flat a bit further, but low end roll-off will be more abrupt (now a 2nd-order filter). None of this actually matters in audio use at frequencies below maybe 20hz.

Also, this has been discussed extensively in many threads with the data posted, there's even an approximation formula to calculate the bump. I'll post the links here when I find those references.

that'd be great and yeah, I saw it moving though it also made a huge difference compared to what the bump was before, see:

I wanna use it on my two bus after the 12 ssl 4k channels and I'm fine with it not being configured for 600 ohm loads since it has to solely work in my environment, I mean, that's the whole point building it, ain't it? ;P

I got three caps before the output trannie currently: 10 + 2.7 + 3.3 mundorf caps in parallel.

 

[emrr]

I don't remember seeing it, sorry if I missed it, did you look at it with a 620-680r load resistance on the output?  Many times that flattens response on amps like this. 

 

[Kingston]

I wanna use it on my two bus after the 12 ssl 4k channels and I'm fine with it not being configured for 600 ohm loads since it has to solely work in my environment, I mean, that's the whole point building it, ain't it? ;P

Yeah, looks good. Time to put on the lid and start using it.

 

[atticmike]

I wanna use it on my two bus after the 12 ssl 4k channels and I'm fine with it not being configured for 600 ohm loads since it has to solely work in my environment, I mean, that's the whole point building it, ain't it? ;P

Yeah, looks good. Time to put on the lid and start using it.

Alright so I did some boosting, just drums now (without toms), been hitting the EQ at -5 dB (dunno how much dBu that is, prolly about 12): 70 Hz and 3 kHz boosts (purple), green's flat.

Here are the files:

https://dl.dropboxusercontent.com/u/70609905/check1.mp3

https://dl.dropboxusercontent.com/u/70609905/check2.mp3

BTW, flat at -16 dB:

Boosted at minus 16 dB vs flat.

For the -16 dB boost I redid the settings so they're not exactly the same as with -5 dB but as close as possible, thought a comparison at different levels might be interesting afterwards.

Hot at -5 dB again with more boost at 70 Hz, is that normal?