Location Tube Mixer

[tomgam1]

Hello everyone. This is my first post here although I've been reading the forum for quite sometime. I am a young Jazz musician who is interested in electronics and recording.

I really like the sound of classic records from the late 50’s and early 60’s. In particular, one of my favorite engineers is  Roy Dunann, who worked for the label Contemporary recording many Jazz greats, including Sonny Rollins, Ornette Coleman, Phineas Newborn and others.

I recently acquired a Nagra portable tape machine and I am going to try to record some of my gigs (and potentially other people) live to two track. The machine I bought is a model III so I designed and built (successfully) tube electronics for it (based on an existing bias oscillator). This recording preamp has a decent amount of gain, maybe 20 dB, and unbalanced inputs.

I intend to design and build a stereo tube mixer to go along with my tape machine.  I will be using high output condenser microphones (m49 style etc). Those, when close miking loud sources such as saxophone, trumpet or bass, output easily 5-10 mV before the step up transformer. In most cases, I see engineers (including real old school ones, like Fred Plaut from Columbia and Rudy Van Gelder) either padding those mikes or modifying their circuitry to match that of the preamps (most of which, in the old days, were to my understanding designed for ribbons and dynamics).

What separated Roy Dunann from the other guys (and part of the reason why his recordings sound so great) is that rather then padding the mikes he ran them directly into constant impedance variable attenuators and then a balanced mixing network, and fed the tape machine from this entirely passive mixing box (there's a few interviews where this is mentioned). The make up gain was all provided by the tape machine, thus making the signal path extremely clean.

Not having access to constant impedance attenuators, I will have to resort to a somewhat different topology.

Here's what I have came up with. Each microphone  (I intend to build a 6 -8 channel mixer) runs into a 1:4 transformer, then a class A stage driving a 50K pot (fader) for volume, followed by another stage driving  a summing network with 50k summing resistors. I chose 12AU7 tubes as from my calculations they would provide adequate gain to have around 1 rms at the output of the busses, and low enough output impedance to drive a 50K pot as well as the mix buss. I would still have some gain in hand from the tape machine electronics. Each channel would draw approximately 10-12mA.

Attached a schematic of what I have in mind.

There aren't many examples of mixers following this kind of topology (other then the Ampex MX10), so I'm looking for critiques, especially regarding the bias point of the tube, the impedance of the mixbusses, and the choice of a 1:4 input transfer as opposed to a more standard 1:10 (to save in cost, and to avoid having too much voltage gain).

Thank you very much.

Tomgam1
Brooklyn, NY

 

[ruffrecords]

If you are using passive mixing you need to ensure unused bus resistors are grounded so you need to turn round your two bus assign switches to do this.

The 12AU7  would not be my first choice of tube for a clean signal path. It is well known to generate relatively high levels of distortion compared to others. With the topology you have chosen I would use a 6CG7 (basically a 6SN7 in a B9A bottle).

Noise performance will depend on the first stage. With 20mV at the transformer secondary and a stage gain of maybe 20dB the first stage output will be 200mV which in my view is too low. You would probably benefit from using a 1:10 input transformer to raise the level at the first stage output to around a volt or so. This stage will be quite capable of driving +30dBu into a 50K load so you will have plenty of headroom.

Cheers

ian

 

[emrr]

What separated Roy Dunann from the other guys (and part of the reason why his recordings sound so great) is that rather then padding the mikes he ran them directly into constant impedance variable attenuators and then a balanced mixing network, and fed the tape machine from this entirely passive mixing box (there's a few interviews where this is mentioned). The make up gain was all provided by the tape machine, thus making the signal path extremely clean.

Not having access to constant impedance attenuators, I will have to resort to a somewhat different topology.

What you propose goes the other way; too much up front gain which will require pads.  You don't need constant Z attenuators, one would argue he used them because they were the common quality option of the day, and one would also argue he loaded the mics too heavily because of it.  1K-2K5 pots would be fine, and there are plenty of passive mixer plans to be found around here that could be applied.  All you need is appropriate make-up gain, which you may have already in the recorder. 

 

[tomgam1]

Hello Ian and Doug.

Thank you very much for your replies. Ian, i have read a lot of the documentation that you made available on your website and it means a lot that you'd take the time to help out. And Doug, I have read countless of your posts about vintage mixing consoles and preamps so I really appreciate your expertise.

I had some musician friends over today and I took some measurements of the Vrms level coming off my microphones. I somewhat underestimated the levels (and this explains why a prototype of a tube microphone that I build a few months ago would clip my converters so easily).

A guitar amp miked at a reasonable distance can yield to an output of easily 10mV (not even playing too loud). A saxophone or a double bass also miked at  a reasonable distance can easily output 20mV, drums even more (depending on what they're playing). This somewhat changes the considerations regarding the gain staging. The input levels that I am expecting fall between 10 and 100 mV.

I appreciate your suggestion of the 6CG7 tube. I read your paper about the mu follower, and the reference to Morgan Jones' work, and it does sound like a much better choice.

I will update the schematic (potentially redraw the load lines all though it'd probably work with the current bias points). If the gain needed is less than what I originally anticipated, I guess an option could be to get rid of the cathode bypass capacitor, although that would impact the output impedance of the stage. Ian, what do you think?

I have drawn a sketch of a passive mixer "Roy Dunann" style, using potentiometers. If I understand this correctly, an arrangement like this, with the bus impedance set at 200R, would cause a loss factor of around 50. The mix buss would then be operating at a level of tens if not units of mV, and I'm not sure if the noise floor would then be acceptable.Do you have any thoughts on this?

If this configuration was used, one would then connect the mixbusses to a step up transformer, then to the tape machine. The active gain needed from the tape machine in this case would be around  10-15. To be honest, I was going back and listening to some of the stuff the Roy recorded on the "quiet" side of things, and I can't help it but to notice a pretty substantial amount of noise(hard to tell what caused it, really, but still). This method might be great if one is exclusively recording relatively loud bands playing loud tunes.

Thanks again and talk to you soon.
Tommaso

 

[ruffrecords]

I would be tempted to change the dual pot value to 1K. This will provide a more 'normal' input impedance and reduce interaction between controls. Unfortunately dual 1K LOG pots are both rare and expensive.

Bus loss is 34dB. If the typical input level is 100mV (about -16dBu) the overall output level is around -50dBu. Even with a perfect mic pre raising this to 0dBu will raise the noise to about -80dBu. A reasonably good tube mic pre will achieve -70dBu to -75dBu. An average one will achieve -60dBU. This is for a single signal so with several inputs going at once the output level will comfortably exceed 100mV so the S/N will improve. On the other hand, you really want the noise from the mixer to be 10dB below the tape noise from the Nagra. Most Nagras will acieve better than 70dB S/N at 15 ips so mixer noise really is crucial.

Cheers

ian

 

[tonzauber]

Tom, a quick and stupid question:

How do you intend to record your stereo feed onto a mono tape recorder such as the mentioned Nagra III?
Best,

Georg

 

[tomgam1]

Hi Georg.

Not a stupid question at all; not being able to afford a Nagra IV-S, I got a hold of a model III for a good price; I then got a hold of a low inductance 1/4 " stereo head with dimensions that would fit the Nagra. Then, I experimented a bunch and I came up with tube electronics to go along with this stereo head that would work satisfactorily (I think the sound is actually much better then then built electronics of the Nagra, as it uses early Germanium transistors, which I dont love the sound of). Essentially Im using my model III as a transport. Attached is a picture of the tape preamp boards (I'm waiting on the enclosure).

Regarding the mixer project: I think all things considered, especially the noise floor (as Ian illustrated), the passive mixer solution doesn't really make sense; to be honest I'm not sure how Roy Dunann made it work so well (I do have a guess: the tape stock they were using had probably way worse performance than what we have today, so the passing mixing board might not have been an issue).

So I went on to building a small prototype of the design I attached earlier, with the only modification being the second cathode is not bypassed (attached is a photo). I am using 100K buss feed resistors and a buss impedance resistor of 10K (approximately). Results seem encouraging although I havent gotten to measuring THD and frequency response and to simulating the condition of the preamp feeding both busses.

I will keep you updated.

 

[tomgam1]

Prototype with Cinemag input transformer

 

[tomgam1]

Hello everyone,
quick update on my project.

I was getting too hot of a signal out the mixer with the first topology I had proposed (transformer-triode-fader-triode-mixing net), so after I few tests I settled on the topology attached (transformer-triode-fader-mixing net), essentially "a la" Ampex MX35. Attached is a schematic. The results of my prototype seem encouraging (I have 2 channels working right now).

Any thoughts on point to point vs pcb for a project like this? And has anyone here used one of those Hammond sloped enclosures?

Thanks!

 

[tomgam1]

Prototype 2 channels (1 cinemag, 1 cheap input transformer)

 

[ruffrecords]

Hello everyone,
quick update on my project.

I was getting too hot of a signal out the mixer with the first topology I had proposed (transformer-triode-fader-triode-mixing net), so after I few tests I settled on the topology attached (transformer-triode-fader-mixing net), essentially "a la" Ampex MX35. Attached is a schematic. The results of my prototype seem encouraging (I have 2 channels working right now).

Any thoughts on point to point vs pcb for a project like this? And has anyone here used one of those Hammond sloped enclosures?

Thanks!

Looks basically OK to me. Not too much gain, sensible tube current and fader values.

I used to do point to point in the early days but now I use only PCBs. PCB design software is free and easy to use. Loads of CHinese suppliers will make you small PCBs at low cost plus you can use a 0V plane which helps screen things.

I have a Hammond sloping box that I never used. Not sure if it is big enough for what you need. If you live on my side of the pond I will happily give it to you.

Cheers

Ian

 

[tomgam1]

Hello everyone,
my apologies for not updating you on my project in a while. Here's a progress report.

I finally put my tape machine electronics in an enclosure. That took a while as many boards are involved and I had to redo the grounding a few times to achieve optimum performance. I also designed and build a VU meter driver board and a monitoring headphone amp based on a few opamps, which took a while.
Now it's working. I will be ordering new tubes and that will hopefully take care of the slight hiss I still have. The S/N I was able to achieve with some old tape is around 65 dB (including the playback electronics). Hopefully it will improve with newer tubes, and modern tape.

Out of curiosity, when placing my last order at mouser, I ordered a few 1K dual audio taper pots to test out the passive solution discussed earlier. It turns out that the noise caused by the additional line stage is not really an issue.

I think I will proceed with the all passive solution. I will however need to perfect the additional gain stage between the passive mixing network and the recording amp. Right now I'm using a 12AY7 that I had laying around in a simple plate follower configuration , but I am open to any suggestions.

Attached pictures of the tape recording amp and of my test setup.

 

[tomgam1]

test setup

 

[L´Andratté]

Hey Tomgam, any updates on this very nice project?

 

[tomgam1]

Hello everyone. My apologies for not updating you more frequently. A few things have happened since I last posted.

After some tests and  after giving it a lot of thought I realized that using the Nagra III transport with my gain stage/recording amplifier in the separate enclosure was a pretty bad idea (Nagra III only takes 7" reels, setup is bulky, bias oscillator module was giving unsatisfactory performance, etc).

So I acquired a beautiful Studer/Revox machine, the PR99, I recapped it/restored it, and I built a tube circuit inside it.
It is comprised of two stages per channel: a gain stage followed by a volume control, followed by a high frequency boost stage (IEC curve, non selectable as I only plan on recording at 15ips). I ended up using the solid state tape head driver from the Revox. It'j just a lot easier this way. Not entirely tube, as there is one transistor pair per channel in the signal path, but the sound is great from what I can tell.

Then, I got a hold of 4 1k audio stereo pots and started building a prototype of a basic passive mixer. I am waiting for the input transformers in order to being able to test the whole rig.

Attached is a picture of the Studer/Revox.

You can see the power supply for the tube section (B+ and dc filaments) mounted under the power inlet. The tube stage in on the bottom.

Let me know if you have any questions.

 

[ruffrecords]

The PR99 is a very good machine. I love the long pole butterfly heads Revox use which  avoids low frequency head bump. They have a wonderful bass response.

Cheers

Ian

 

[jdurango]

If you are using passive mixing you need to ensure unused bus resistors are grounded so you need to turn round your two bus assign switches to do this.

The 12AU7  would not be my first choice of tube for a clean signal path. It is well known to generate relatively high levels of distortion compared to others. With the topology you have chosen I would use a 6CG7 (basically a 6SN7 in a B9A bottle).

Noise performance will depend on the first stage. With 20mV at the transformer secondary and a stage gain of maybe 20dB the first stage output will be 200mV which in my view is too low. You would probably benefit from using a 1:10 input transformer to raise the level at the first stage output to around a volt or so. This stage will be quite capable of driving +30dBu into a 50K load so you will have plenty of headroom.

Cheers

ian

Aloha Ian,

I know you are obviously more than capable of building your own console from scratch, but it sounds like you're very close to an Ampex MX10/MX35 to begin with. They can be linked together to achieve an 8x2 configuration if need be. It would be easy enough to mod the LCR pan with L-LC-C-RC-R or a variable pot if need been. Beyer xformers can be "upgraded" easily. d

Just a thought. I was looking for something very similar a while back and ended up with an MX10 that I've been very happy with (run into an Ampex 354)....then again, I couldn't build myself a custom console if my life depended on it

Good luck!

 

[Manunagra5]

Hi Georg.

Not a stupid question at all; not being able to afford a Nagra IV-S, I got a hold of a model III for a good price; I then got a hold of a low inductance 1/4 " stereo head with dimensions that would fit the Nagra. Then, I experimented a bunch and I came up with tube electronics to go along with this stereo head that would work satisfactorily (I think the sound is actually much better then then built electronics of the Nagra, as it uses early Germanium transistors, which I dont love the sound of). Essentially Im using my model III as a transport. Attached is a picture of the tape preamp boards (I'm waiting on the enclosure).

Regarding the mixer project: I think all things considered, especially the noise floor (as Ian illustrated), the passive mixer solution doesn't really make sense; to be honest I'm not sure how Roy Dunann made it work so well (I do have a guess: the tape stock they were using had probably way worse performance than what we have today, so the passing mixing board might not have been an issue).

So I went on to building a small prototype of the design I attached earlier, with the only modification being the second cathode is not bypassed (attached is a photo). I am using 100K buss feed resistors and a buss impedance resistor of 10K (approximately). Results seem encouraging although I havent gotten to measuring THD and frequency response and to simulating the condition of the preamp feeding both busses.

I will keep you updated.

Hi Tom, i'm new here and i registered this french morning after reading your old post as i'm fan of Roy like you. I'm not sure of that but i think that U47 and C12 microphone had a condiderably hotter output level in their early days. As you know, Roy put a PAD in each input of his passive console that was built for ribbons. I read that Neumann (and AkG), after a lot of demands in USA, introduced a PAD directly in the Power supply of their mics to make american sound engineer lifes easier. Maybe you didn't succeed in your project due to relatively low levels outputs from contemporary microphones. But as i said, i'm not sure. I would like to know how much was the gain of "for ribbons" mixing consoles. With this data, i think we could have a good approximation of first German mics output levels...
Manu

 

[Bill Wilson]

You are correct, had 2 u47's & wired them for 50 ohm output. This lowered the output to a more reasonable level. Got mine direct from Germany; courtesy of my friends fiance who purchased the mike bottles & 30' cable w shock mount, total cost $175.00 ea. Made my own 105vdc power supplies