THAT1570/5171 DIP Adapter

[JR.]

I ran some measurements and with 150R input termination I get -127.5dB EIN @60dB gain 20-20k BW. With inputs shorted I get -130.8dB EIN @60dB gain, and -131.5dB EIN @75dB. Generally speaking, could I see EIN improvement with swapping my 5532 with something more high-end? What other factors would improve my EIN figure by 1-2dB?

Thank you!

Perhaps cooling it down to absolute zero....

The difference between input shorted at 60dB gain and 75dB gain, is the value of the Rg resistor. For noise analysis this gain resistor can be considered in series with the input. The difference between 150 ohm termination and input shorted is obviously the 150 ohms. What this tells me is that you are pretty close to optimal.

You don't cite the noise with 150 ohm and 75dB, but i will predict you don't see the 0.7dB drop you saw with input shorted because the noise of the 150 ohm resistor will dominate (adding a small amount to a larger amount changes the sum of the two a small fraction). With input shorted you are seeing the difference between Rg(60) and Rg(75), with 150 ohm termination the comparison is 150+Rg(60), vs 150+Rg(75) so a much smaller fractional difference.

Messing with active devices at the end of the chain will make almost no difference to noise unless they are very bad. The reason is because their noise contribution is after the large gain stage, so the input noise floor times the gain is already elevated (by 60 dB) so your ein numbers translate to "actual" noise at the output driver of -127 dB + 60dB so -67dB (u or V, or?). The noise of a 5532 running at modest gain has little problem being way quieter than -67 dB.

The late stage devices may make more difference at low gain, when the input noise is boosted less, but you will also get more noise contribution from large Rg values.

Note: there is a couple dB difference between dBu and dBV,

JR

[mediatechnology]

The same is true for the output figure +26.6dBu running with +/-15V supplies, which is almost 47Vpp. I'm sure I don't know how to calculate this properly, could you please give me a hint?

I believe that the input and output maximum levels are differentially-measured and thus 6 dB higher than the single-ended levels that you're used to seeing.

[promixe]

Thank you Wayne and JR for your answers. I have another question, perhaps risking to be burned on fire or killed with stones, but here we go.. =))

I'd like to implement a switchable input transformer PURELY as an end-user ARTISTIC feature, because otherwise it makes no sense at all. =) The trick is: the transformer should be able to switch in/out while phantom is powering a mic. It would be much simpler to figure out otherwise (in MCU), i.e. shut off phantom, discharge dc coupling caps, switch transformer in, turn phantom on, and similar for the other way around. But I want the end-user to be able to A/B transformer/ac-coupled on the fly and I don't mind moderate clicks while switching between tx in/out, it's a time gap that I'd like to get rid of. Therefore I'm thinking something along these lines:



I think phantom-wise this circuit is OK, but I'm not sure about impedances when RE6 is switched to the "capacitor mode", because it has the transformer primary (84ohm DCR when in series) sitting in the circuit...? Is it in parallel with the rest of the things? (R1+R2, R7+R8, R11+R12) Then it surely wouldn't work.

I have another version of the circuit where there is an extra relay right past RE3 (polarity switch) switching discretely between transformer and capacitor paths, but then when the TX is switched in C5 and C6 are going to discharge through R11 and R12 and when switched back into "capacitor mode" there will be a big glitch while the caps charge back up. To keep the caps charged up while in "transformer mode" I have to provide 48V to both paths at the same time via R31 and R32...



I know it's kind of a mess, but if you could point me in the right direction and/or tell me which scenario is closer to being somewhat hopeful I'd really appreciate it. =)

Thank you!

[mediatechnology]

Thank you Wayne and JR for your answers. I have another question, perhaps risking to be burned on fire or killed with stones, but here we go.. =))

I think you may mistake us for some other place

I think RE6B is mis-drawn WRT the transformer secondary in the figures. You're really going to have to switch both legs or the common mode rejection may be upset due to stray capacitance. When you lay all of this out be sure to maintain physical as well as electrical symmetry.

I think the first example is worth a try even though the transformer primary still loads the source when switched to AC-coupled mode. (That doesn't seem any different to me than a bridging "split" when preamp inputs are paralleled.) It would be unsuitable for low impedance unbalanced inputs unless the opposing input is grounded because otherwise the signal would appear essentially in common mode. An slight advantage of having the primary always in-circuit is that the DCR of the primary will improve equalization of the DC differential mode potentials as phantom is turned on and off. (For differential phantom faults the primary DCR will steer current away from the preamp IC.)

Figure 2 is probably the cleanest solution overall WRT the absence of added source loading and ought to work well.

The problem with both figures is that the transformer secondary is loaded with 2K differentially by the bias resistors. (R7 and R?) That may or may not be a problem depending if there's impedance transformation due to transformer ratio. If it's 1:1 2K might be an OK load. To remedy this with higher ratios you might consider T-bias where you could go to 20K or more differential load.

There may be other issues I haven't noticed yet that JR will be much quicker to pickup on. I haven't even finished my first cup yet.

[mediatechnology]

OK, just had some more coffee. T-bias isn't going to raise the differential resistance, just the common mode which doesn't help the transformer loading.

[JR.]

Once again I am momentarily frozen by the question why? Or exactly why that (1:1)..?

If trying to mimic classic transformer mic preamp sound, it seems you want a step up transformer not a simple 1:1 to fully capture the characteristic transfer function. Parsing out all the sundry influences of transformers on preamp sound is more than I can or want to fully explore right now (i already burned too much time just logging in...long and boring story), but my gut screams that just throwing a 1:1 in series is not the same.

The next question becomes if we use a proper mic pre transformer, what to do with the extra signal voltage? The transformer secondary with step up expects a higher impedance loading termination, than a low z mic preamp input, so perhaps a simple resistive pad would keep both sides happy there.

There are several details to work out to the right of switches in your truncated schematic, but I'd rather not hash out specific details of your program until you convince me this is really where you plan to go.

If you have relays burning a hole in your pocket, something else to perhaps consider is shunting across the DC blocking capacitors when phantom power is not called for.. If there is no DC to block, why put those bad actors in series with your signal path.. This too probably has a number of unintended details to work out, but offers a purer signal path, at least in theory.

Of course adding a transformer suggests movement in the exact opposite direction. I only suggest that if you add transformer coloration on purpose add coloration that will sound more like a typical mic preamp. I don't know that transformer sound is that generic or universal.

of course i could be wrong...

JR

PS: I have a passing interest in cybernetic control of analog audio circuits. Not unlike this project. I have (too many) ideas for computer controlled comp/limiter, and other projects associated with simple level and pans... I would love some advice about how to get the last mile between the users mouse clicks and a serial digital data stream I can decode for specific analog controls. I am guessing something like midi outputs, a variant of serial com that can be decoded (for simple level and pan). The uber comp may need some more sophisticated control layers (custom sliders?).

This is not an immediate project for me, so no hurry, but i would love to hear advice from you, who are already doing something similar. My brain is already crushed by too many secret handshakes so I don't want to have to design PC interfaces, or PC software too.

[promixe]

I think RE6B is mis-drawn WRT the transformer secondary in the figures. You're really going to have to switch both legs or the common mode rejection may be upset due to stray capacitance. When you lay all of this out be sure to maintain physical as well as electrical symmetry.

Yes, thanks for pointing this out - this is a drawing mistake, as I meant to hook it up symmetrical (i.e. secondary negative to pin 5 of RE6B).

The problem with both figures is that the transformer secondary is loaded with 2K differentially by the bias resistors. (R7 and R?) That may or may not be a problem depending if there's impedance transformation due to transformer ratio. If it's 1:1 2K might be an OK load.

I was thinking 1:1.25 transformer ratio with 1k5 bias resistors initially. The mic would then see 2k15 load and the 1570 would see about 500R, which is near ideal. Then I calculated that even with 1k bias resistors impedances would still be suitable (mic would see 1k5 load and 1570 would see ~450R as source). So, something in between 1k2 and 1k5 bias resistors would be suitable, I think, for 1:1.25 TX.

The downside to figure 2, I think, is when it is switched to "transformer path" the phantom network draws twice as much current per channel (up to 28mA instead of 14mA). But if I can switch between the paths without crazy audible artifacts and not blow up anything while switching this current draw is fine with me. =)

Figure 1, if it can work, would be better because no extra relay is used, DC equalization, as you pointed out, and phantom network uses less parts and draws current as usual. I just got confused by how to measure impedances correctly with the transformer primary hanging in the path permanently. If I were to draw an equivalent circuit for quicker analysis would I replace the TX primary with a 84R resistor sitting in between audio + and -?

Thanks!

[mediatechnology]

If trying to mimic classic transformer mic preamp sound, it seems you want a step up transformer not a simple 1:1 to fully capture the characteristic transfer function.

I agree. And as you also point out then you have this added gain and high impedance loading requirement. At that point the 1570 isn't the best preamp option.

For a switchable "transformer sound" option you might consider switching an output transformer in the path. That's what Roger has done with the pico on some builds.

[promixe]

Once again I am momentarily frozen by the question why? Or exactly why that (1:1)..?

Yep, that exact question temporarily paralyzes most engineers, and immediately excites nearly all musicians/producers. =)) I'm more of the latter crowd myself, so I think it would be a neat feature to have a super-clean preamp that could morph into a more vintage, more "analog" (NOT in an engineering sense) sounding with push of a button. =)

If trying to mimic classic transformer mic preamp sound, it seems you want a step up transformer not a simple 1:1 to fully capture the characteristic transfer function.

Yes, I completely agree, unfortunately there aren't many transformer options that are over 1:1 but less than 1:2 (1:2 seems already too much if bias resistors are to be kept low). Although people have done a similar (not switchable) thing with 1:8 transformers into THAT1510. I'm not sure how it's done.

Ideally I would want a small core transformer, so it saturates a bit more at higher input levels and with something like 1:1.25 that's got to give the preamp a distinct sound, not like a clean 1:1 isolation transformer.
It's funny, but I did experiment with Carnhill VTB9045 (Neve mic input) transformer wired as 1:1 in front of my 1570 preamp and the difference in sound was a lot more subtle than what I expected. It had a slight bass roll-off, primarily, I think, due to non-ideal loading of the secondary, otherwise it might have had no perceivable difference at all...

The next question becomes if we use a proper mic pre transformer, what to do with the extra signal voltage? The transformer secondary with step up expects a higher impedance loading termination, than a low z mic preamp input, so perhaps a simple resistive pad would keep both sides happy there.

So, something like an H-Pad right past the secondary but before RE6 could take care of higher load impedance in a higher ratio transformer?

There are several details to work out to the right of switches in your truncated schematic, but I'd rather not hash out specific details of your program until you convince me this is really where you plan to go.

Yes, this is where I'm going. =) As I mentioned above, it has an appeal to end-users and if done right (primarily in "the way it sounds" sense) it would be a useful producer/recording-engineer tool. Millenia STT-1 has it, but it wasn't done right, judging by the way it sounds. Either they cheaped-out on the transformer or over-did the "vintage" sound. It sounds muffled, choked, with lost high-end, etc. So, I think there is an opportunity to explore, especially now that everybody is crazy about "warming-up" their ultra-clean digital setup. =)

PS: I have a passing interest in cybernetic control of analog audio circuits. Not unlike this project. I have (too many) ideas for computer controlled comp/limiter, and other projects associated with simple level and pans... I would love some advice about how to get the last mile between the users mouse clicks and a serial digital data stream I can decode for specific analog controls. I am guessing something like midi outputs, a variant of serial com that can be decoded (for simple level and pan). The uber comp may need some more sophisticated control layers (custom sliders?).

This is not an immediate project for me, so no hurry, but i would love to hear advice from you, who are already doing something similar. My brain is already crushed by too many secret handshakes so I don't want to have to design PC interfaces, or PC software too.

I thought about this a lot and what's most appealing to me personally is going the WEB route. Pretty much every computer device on the market (including phones, etc.) has a browser built in. These browsers are optimized for HTML5/Java speed and are very stable. It's a lot easier to code an interactive web page that IMMEDIATELY works on every platform/CPU/device rather then coding a cross-platform application code, maintaining it is another story. You use a auto discovery protocol, something like Bonjour to automatically assign your audio device a domain name, you go to 1.jrdevice.com and there is your control interface for it. There is no serial-to-USB or serial-to-Ethernet adapter you need to plug in, in fact, you can plug the audio device into any router or even control it being in a different country, while listening to the changes through ISDN stream. =) On the audio device side you run a tiny embedded web server with your web pages loaded in and the server translates what it gets from the client into SPI. Hardware-wise you'd need something like PIC32 or AVR32 chip with Ethernet-MAC built in + a separate Ethernet PHY chip.

SSL is doing it like that with their consoles, primarily for preset/recall management, configuration, etc.

This may seem too complex of an idea, but the most work would be on the audio device side, while on the user side there wouldn't be much to do because you would just use existing technologies (TCP/IP, browser, HTML5/AJAX). Now, if you needed absolute precision or sample-accurate control of your device then this idea wouldn't work as there is a typical 3-6ms delay between push of a button on the screen and SPI message out of the chip (given typical home LAN latency). Again, this is what seems ideal to me personally, because it looks daunting to code an application for OSX, getting it to work right, then doing a Windows version (even if reusing half the code, which then would compromise Mac's Objective-C workflow), then Linux, iPad crowds get left in the cold, then next version of Windows comes out and here I am redoing the app because some API I relied on is now discontinued. =) Then there is extra hardware on the client side that needs to deal with RS-232/RS-485 (something like a USB adapter, that you forget at home when rushing to an important remote recording gig). =) Don't discount the benefit of wireless when going TCP/WEB route. I was playing a concert a few months ago, the sound guy was walking around the venue with an iPad adjusting everything on the fly with his Yamaha setup.

An RTAS/VST/AU plugin could be coded that would run inside of a DAW and maintain TCP/IP stream into your device. Now we're talking about full automation and recall of your analog gear within your Pro Tools session file - a dream for mixing / recording engineers.

[JR.]

Once again I am momentarily frozen by the question why? Or exactly why that (1:1)..?

Yep, that exact question temporarily paralyzes most engineers, and immediately excites nearly all musicians/producers. =)) I'm more of the latter crowd myself, so I think it would be a neat feature to have a super-clean preamp that could morph into a more vintage, more "analog" (NOT in an engineering sense) sounding with push of a button. =)

Yup, that's me no imagination...

If trying to mimic classic transformer mic preamp sound, it seems you want a step up transformer not a simple 1:1 to fully capture the characteristic transfer function.

Yes, I completely agree, unfortunately there aren't many transformer options that are over 1:1 but less than 1:2 (1:2 seems already too much if bias resistors are to be kept low). Although people have done a similar (not switchable) thing with 1:8 transformers into THAT1510. I'm not sure how it's done.

Ideally I would want a small core transformer, so it saturates a bit more at higher input levels and with something like 1:1.25 that's got to give the preamp a distinct sound, not like a clean 1:1 isolation transformer.
It's funny, but I did experiment with Carnhill VTB9045 (Neve mic input) transformer wired as 1:1 in front of my 1570 preamp and the difference in sound was a lot more subtle than what I expected. It had a slight bass roll-off, primarily, I think, due to non-ideal loading of the secondary, otherwise it might have had no perceivable difference at all...

I am not the right person to tweak transformers as sound effects.

Back in the '70-'80 when I designed real studio delay based effects the 'effects" were not very sublte.

Since then I have had several experiences with the fringes (more tube than transformers) but in both cases the results were not simple or linear.

In one case i had a pile of slow moving inventory to move... An actual 4 input 2 output tube line mixer (designed for guitar efx). I figured it would be a no brainer to sell these to the phools, for the tube sound .. no takers. I even increased the gain to make it easier to clip. But the peavey badge was a bit of downer.

incident # two was a very nice tube based mic preamp that I consulted on as a friend of the court. I had the real tube engineer add a header/jumper to the circuit that allowed us to reduce the amount of open loop gain to negative feedback margin, so i could make it less linear and express more of the tube sound and less of the precise negative feedback network. This hidden feature was kept secret in reserve if anybody complained that it was too clean (it was clean). Not one customer asked for it to sound more tube like so the header was permanently hard wired in the clean position.

So I don't trust my judgement for subtle efx, and philosophically I leave that for others.

The next question becomes if we use a proper mic pre transformer, what to do with the extra signal voltage? The transformer secondary with step up expects a higher impedance loading termination, than a low z mic preamp input, so perhaps a simple resistive pad would keep both sides happy there.

So, something like an H-Pad right past the secondary but before RE6 could take care of higher load impedance in a higher ratio transformer?

Thinking about this some more, the transformer impedance ratio is the square of the voltage ratio so a simple pad doesn't get it done simply...
never mind... try it your way.

There are several details to work out to the right of switches in your truncated schematic, but I'd rather not hash out specific details of your program until you convince me this is really where you plan to go.

Yes, this is where I'm going. =) As I mentioned above, it has an appeal to end-users and if done right (primarily in "the way it sounds" sense) it would be a useful producer/recording-engineer tool. Millenia STT-1 has it, but it wasn't done right, judging by the way it sounds. Either they cheaped-out on the transformer or over-did the "vintage" sound. It sounds muffled, choked, with lost high-end, etc. So, I think there is an opportunity to explore, especially now that everybody is crazy about "warming-up" their ultra-clean digital setup. =)

part of my resistance is the fashion aspect of colorations... I find it hard enough to make the linear path friggin perfect.

JR

PS: I have moved my cyber veer to a different post in case we want to make this a different thread... i am too guilty of veering off on tangents.