Another PRR Vari-Mu question...

[bluebird]

The 12Au7A is built a little better I think. It really doesn't matter. what matters the most is brand. I've heard the best ones are Amperex. the ones made in Holland. All of them can be microphonic. I don't know which of the newer brands are good...

 

[Svart]

thanks bluebird, I'll get them, at least i can then start on this.

:guinness:

 

[NewYorkDave]

The 12AU7A is just a lower-microphonic version of the 12AU7.

 

[PRR]

> is there any difference between the 12au7 and the 12au7A?

Back in 1959, the -A version was "better" in some way. Sometimes built by the best workers, or better materials, or extra testing. In a few cases, the -A is rated for more abuse than the non-A version. (A 6L6GC will take a LOT more abuse than an original 6L6G.) And sometimes the same as the no-A version, just priced higher.

Back then, if you NEEDED a best-possible tube, you bought the -A versions. A radio station with a dead tube didn't have time to test several tubes to find one with low noise or microphonics, so they just bought the Premium versions.

As far as function goes: all the same for this project. No abuse involved. Low noise and low microphonics would be nice, but "Premium" tubes may have aged poorly, and some non-Premium tubes were premium quality. In DIY you can try several tubes, unlike on-air broadcast where it HAS to work NOW.

Unless you know your eBay seller, I would get common tubes from http://thetubestore.com/ Good service and good tubes. Argh, the dull 12AU7 has become "hip" and there are too many "sexy" versions.
http://thetubestore.com/12au7types.html

If you are poor, get the $9 ElectroHarmonix. I hope it is a good copy of the classic 12AU7. EH does market decent tubes.

For $13-$20 you can get classic US old-production 12AU7s and their military equivalents. These are the tubes I based my sims and calculations on.

Don't buy generic Chinese "12AU7" except for experimentation. Small changes in grid-wire layout may work fine in guitar amps but have less (or more!) of the vari-gain bend we need.

Don't pay $50 for sexy-brand 12AU7, unless you really want to, and have proven basic operation with a $10-$20 tube. The Amperex is a fine tube, but so is the GE or Philips or Raytheon.

 

[NewYorkDave]

In addition to the above, the "A" suffix sometimes indicates a controlled heater warm-up characteristic.

 

[kent]

I tried a bunch of different ones in mine. An old Telefunken sounded best but the sides didn't match very well. I ended up with some EH's. Best bang for the buck for me in this project.

cheers,
kent

 

[Svart]

ok, Tubestore it is. I deceided not to buy those ebay tubes at the last minute anyway.. good thing i did!


thanks everyone!

 

[Svart]

ok got this thing built up and preliminary tests show something weird with the PS.. +12=ok, +100=ok, -16=ok...+1.5 shows -1.5..? also +16 shows -8.5.. any ideas? haven't tried it yet though..

 

[PRR]

> +16 shows -8.5..

There's your problem. Re-re-check the wiring around the rectifier and caps. Might be a missing common wire.

 

[Svart]

http://www.conditionedresponse.com/DIY/C5/C5PCB.pdf

that's the board I used.. It *looks* right unless i am just not seeing something, maybe I'm looking too closely.

I don't understand how the +12 and the -16 can be ok but the +16 and +1.5 are not since they are from the same rectifier... The etch turned out beautiful with not a trace of a problem(pun intended.. :green: ).

I'll triple check everything again tommorrow when i get a chance, but I've already double checked the resistors and the caps on the ps and all turn out just fine. the -8.5 is with the ICs in place or not, same for the -1.5..

 

[PRR]

> http://www.conditionedresponse.com/DIY/C5/C5PCB.pdf ...that's the board I used..

Hmmmm, I had not seen that before. Thanks. Someone did think-hard about some of the little-bits I omitted. I would do the grounding different, but that does not explain your problem.

> It *looks* right

I see this error which exactly matches your problem:

Fortunately it is a part-value problem, not a board problem. If you hate to extract the erroneous 10K resistor, just tack a 220R resistor across the 10K.

I'm assuming you are getting +18V to +12V (unloaded and loaded) at the "+12V" heater-power output (you never bothered to mention that, or if the LED lights {it should}). The 10K/220R mix-up sure explains what happened to the +16V, and perhaps the +1.6V (too lazy to see where it goes).

 

[Svart]

yes, sorry the unloaded voltage for the +12 is hovering around +18, loaded it's right at 12.2.

the LED does Light which was partially confusing me as to why the meter would show negative voltage when the LED is properly biased and lights up.. this leads me to my next thought:

I'm also starting to think that my meter might be acting funny. It's a new Fluke 110. I actually think the auto sensing is a little buggy, I'm going to check the voltages again before and after the resistor change with my old meterman meter, one that is old as dirt but I know is dead accurate.

changing the resistor is no problem, I actually pulled them out to double check their values the other day. Now that you mention it I see that the resistor should have been different but I've been so flustered with work that I haven't had more than a few minutes to *work* on this per day and just plain overlooked it! Hey but this is why I posted this on here, to get other people to look at it a different way.

thanks again PRR, I'll let you know how it goes, If the voltages look good then I should be able to fire this up this weekend and give it a try!

:thumb:

 

[kent]

I see this error which exactly matches your problem

Thanks for the heads up. I'll fix that soon.

cheers,
kent

 

[Svart]

ok changed to 220 from 10k and everything is good to go!

will be trying this thing out this weekend!

:thumb:

 

[PRR]

> the LED does Light

How about now that you have positive voltage? The plan is right, but if it lit before then it seems you must have the LED in backward. It really has to light (dimly).

One reason I never got around to drawing a 1.6V bias supply is that it may need a lot of current, and I never sat down and figured it. It might only be AC current, so the 1,000uFd cap shown might stabilize it fine. Also it might not be 1.6V for a real 12AU7.... I shouldn't have trusted my model to be that precise. It should work as that plan shows, but the output limit may not be 2.0-2.8V peak and it might suck-down the 1.6V reference in very-heavy limiting. But try it and see.

 

[Svart]

yes I did have the LED in backwards once I placed the correct resistor in and the bias changed. strange, but I now see how it happened and why.

the LED I am using is right around 1.55v, the closest one i could find out of the bunch i had lying around.

Now, just taking a quick look at the schematic again, how much fudge room do i have to increase the output of the 5532(maybe to +4 since that is what I'm running..) without seriously changing the charachteristics of the sidechain? I would like to try changing the feedback resistor and changing the 1k pot(r39) to something like a 5k for a little more fine tuning with a higher output. I'm not sure what R40 does, is it to limit the lowermost end of the adjustment?

just some thoughts I'd like to play around with. if you have some insite, please share!

 

[ebartlet]

[quote author="Svart"]Now, just taking a quick look at the schematic again, how much fudge room do i have to increase the output of the 5532(maybe to +4 since that is what I'm running..) without seriously changing the charachteristics of the sidechain? I would like to try changing the feedback resistor and changing the 1k pot(r39) to something like a 5k for a little more fine tuning with a higher output.
[/quote]

I think I can answer this....

Changing R17 (feedback) to 10k, with a 270 to ground should give you an additional 12db of gain which would put you in the ballpark for +4.

Obviously you will need to attenuate this for you sidechain, so it still operates in range. A 1.8k resistor between the pot wiper and C8, and then a 1k between where 1.8k meets C8 to ground, will attenuate 12db

Keep R39 a 1k

Please post some clips if you can, I wanna build one sometime and would like to hear this thing. PM me if you need me to host em...

 

[Svart]

thanks for the tips! i'll be trying them for sure! yeah I should have some clips that will needs hosting too. thanks!

 

[Svart]

testing the PRR vari-mu right now, the compression is very smooth and almost unnoticable. the circuit has a good sound so far, the tubes add a small bit of warmth to the sound which i'm really digging! so far we've done some acoustic guitar and some drums featuring the WM61 panasonic electret capsules with the linkwitz mod and a matched jfet circuit(thanks to marik for this..) and everything is sounding really good! more tinkering to be done tonight, but so far it's great!

I'll have some clips after the weekend.

:thumb:

 

[Svart]

ok another update.

I did a moderate amount of testing and modding over the weekend. Here's the rundown:

The input as designed really only likes seeing a balanced input. cold (- pin) shorted to ground gives VERY noisy output (noise floor around -40db@unity!, mostly 60hz hum) while leaving -pin open gives almost nothing.

I decided to try a regulated supply for the opamps and added +-15 Vregs where the 220Rs would go. no real change in sound/noise. additional .1uf polyester caps on all rails didn't seem to do much for the noisefloor or sound quality. I did not test this theory with a scope, jsut with my ears! I also didn't try adding a 12v reg to the 12v supply either.


60hz hum is only present with the tubes in place. without them, the noise floor drops to -85db, the limit of my ability to test, composed mainly of white noise.

balanced input gives around -65 to -70db noisefloor with faders @unity. not too shabby.

I have a sample, I still need to convert to mp3 when i get home from work, I'll send it to my gracious host tonight, however the compression is very UNnoticeable. you can tell that it's compressing when you push the faders up and watch the VU on your mixdown deck and hear it starting to pump at higher compression, but it's very subtle!

oh and the cheapo radioshack transformers are fairly UNcolored and don't seem to affect frequency response much at all to my ears, again I didn't scope it so i don't have a clinical analysis to back this up.

I did the +4 mod suggested earlier in this thread and it works like a charm, maybe a slight bit hot though, but definatly within a few % of the goal.

:thumb:

EDIT: I really need a balanced output too.. I assume there will be no problem with running the outputs to a pair of 5532s set to unity right? the signal going to one half of the 5532 to the +, the signal also going to the other half going to the - and the outputs becoming my + and - signals for each channel.. at least this is the plan until i can proto it and tweak.