Yet another "bright idea...."

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

CurtZHP

Well-known member
Joined
Mar 21, 2005
Messages
634
Location
Allentown, PA
It's been a very slow, uneventful week at work.  You know what they say: "Idle hands are the devil's workshop."  So, here I am!

I found a fistful of Analog Devices SSM2015 chips.  Apparently, these were pretty popular preamp chips in their day.  I started kicking around building something around a couple of them.  Then I started looking at currently available devices, like the THAT 1512.  I guess my last two projects left me kind of "tubed out," so I'm leaning toward building something solid state this time around.

Most of the designs I see for these chips assume a capacitor coupled input, but what's the story with using a transformer?

So far, what I've scratched out on paper is something like this...  Nothing new.  Keeping it relatively simple, but leaving room for something more complicated, maybe....

Input-->(phantom power, maybe a pad)-->transformer-->solid state fixed gain stage-->(selectable HPF)-->level control-->second solid state fixed gain stage-->output transformer-->(polarity switch)-->output.

Trying to decide how much gain to have in each stage.  My math could be wrong, but using an OEP input transformer like the one in the Gyraf G9, I was calculating 16dB of gain just in the transformer.  The output transformer would likely cost me 6dB.  So, then it would be a matter of figuring out how much gain in the active stages would get me to about 50-60dB.  Or should I be looking strictly at voltage gain as opposed to dB?

What about giving the first stage selectable gain (five or six position switch), with the level control handling output level?

For about five minutes, I got really crazy and started thinking about a three band tone control and "single-knob" compressor stage before the final gain stage.  (Those s.o.b.'s at THAT and their design notes!)

Please talk me out of this!
:eek:
 
I just did a pre very close to what you describe. Here's a "simplified" schem (it was a dual):

stp-0.4simplified.pdf

However, a few things about this circuit.

1) It's designed to work from a single 48V supply in which case VGND is virtual and will effectively be at 24V relative to the chassis which is to say, you cannot connect the VGND to the chassis. If you use a proper triple supply you can but you will also want to put the servo caps on the 1646 and maybe also add a coupling cap between the 1510 and 1646. Or use a different output entirely.

2) I used an input transformer only because the enclosure already had little bullet transformers in it but you do not need transformers at all. In fact, the THAT 1510 / 1512 is specifically designed to yield good CMRR performance without a transfomer [Correction: I was thinking of THAT 1206. There's nothing special about THAT 150x wrt CMRR so using a transform would improve CMRR significantly]. You could just do coupling caps and the usual circuit that you'll find in the THAT 1510 / 1512 datasheet.

3) The low cut is a little unusual for a CFP because normally the capacitor in the gain control network is huge to eliminate electrolytic capacitor distortion. However, I decided to try Nichicon UES "MUSE" bipolar 35V and it seems to work fine. I have not detected any measurable distortion with the low cut in vs. out. If you follow the link about the 48V supply, there are links to other posts about using a smaller cap in the CFP gain control network to implement low cut.

Note that the THAT 1510 / 1512 can make a LOT of gain. I had the 1510 making a little over 70dB and it was perfectly stable near as I can tell. I trim the max to around 67dB I think and it works fine. Just in terms of numbers, it's the best performing pre I have.

Note that you would almost certainly NOT want to use a step-up transformer because the 1510 / 1512 is designed for a certain source impedance. So it will actually be noisier with the step-up. If you want to use a transformer, use a 1:1 (preferably with center tap for phantom like my circuit for maximum CMRR performance).
 
Thanks for the tips!

I did a little more digging on the THAT chips, and I concur regarding the transformers.  Diminishing returns at best.
Using coupling caps would certainly be simpler, and cheaper; but I have one concern about that...

In order to maintain CMRR, would those caps need to be very closely matched, and how does one even do that, short of sitting there with a capacitance meter and measuring dozens of them?  I had also read somewhere that using non-polar caps for such a purpose was superior to using polarized?  Thoughts?

Fortunately, I already have a bipolar supply built, complete with adjustable regulators and a 48V phantom rail.  Built it a while back for prototyping stuff like this, and I have a PCB layout for it, so I can duplicate it if needed.
 
CMRR can be affected by electrolytic coupling capacitor mismatch at low frequencies but if you make them bigger, the effect deminishes [1]. But otherwise, generally no, it would not be a problem because their impedance is probably less than 0.5 ohms which is small compared to the impedance of the inputs which is usually 1K-ish. Bipoloar doesn't matter. If you want to be pedantic about it just get something low-esr. Bipolar are known to be low esr. Caps that are "audio" are just low-esr. Higher voltage usually equates to lower esr but with higher leakage which might be important if you have a switched in there like a pad but that's easy to fix with drain resistors and a pad is not really necessary with a THAT pre because it has good performance at low gains.

[1] Although you have to be careful about how you apply phantom power because you are effectively charging those coupling caps in a way that could generate current spikes of several amps. There's a paper on it called "The phamtom menace".
 
CurtZHP said:
In order to maintain CMRR, would those caps need to be very closely matched
Actually my statement about 1510 being designed for good CMRR was wrong. It is better becuase of closely matched parts but I was really thinking of THAT 120x line receiver. The 151x doesn't have any extra special circuitry to help with CMRR like the 120x. So a good transfomer would actually be a nice improvement. Especially if you're running a mic level signal over a long run in an electronically noisy environment.
 
I think I have a pair of 100uF non-polars laying around somewhere I could try.

If I go the transformer route, I'd have to rethink the gain structure to compensate for that, right?
 
Bare in mind they have to be 50+ volts.

If you use a transformer that is not 1:1 then yes, you would need to consider that for your target gain. But that's not how I would pick the transformer. I would pick the transformer to match the optimal source impedance (OSI) of the IC. I don't know what that is because there's nothing in the datasheet about that but if the IC is designed for interfacing with a mic, I would assume that a well shielded 1:1 600 ohm transformer should be used (preferably with a center tap so that phantom can be delivered through it). It might be that the OSI of the 1510/1512 is actually a little higher in which case a small step up might actually shave a dB or so off of the noise floor.
 
The WHOLE point of those SSI and THAT chips is to NOT use a costly heavy and narrowband transformer!

If you use a transformer, you can do a (slightly) better design with a step-up and an amplifier optimized for higher impedances.... so don't use the "transformerless" preamps.

The input caps typically should be *large* to minimize 1/f noise (rumble hiss). If they are large enough for that, they typically have little effect on CMRR. THAT surely covers this design detail in a paper?
 
For actual values of 100u and 90u, CMRR at 50Hz is 49dB.
 

Attachments

  • 100u-90u-CMRR.gif
    100u-90u-CMRR.gif
    12.7 KB · Views: 30
If you want to try a transformer (for the hecking heck of it  ;D) you can always try a 1:1 input transformer from Jensen...the mic splitter transformers come up on Ebay often for very little $.
 
PRR said:
The WHOLE point of those SSI and THAT chips is to NOT use a costly heavy and narrowband transformer!
Maybe you care if you're making cheap 16 channel mixers by the truckload. But this is DIY. An OEP A262A2E is ~$20 USD with the shield and the THAT 1510 is $5 USD. The transformer provides great CMRR and totally solves "The Phantom Menace" problem and the 1510 provides great low noise performance at a wide range of gains in a single easy to use package. My time is worth much more than the extra cost of these parts. So I'd rather just use whatever parts are going to make a simple build with great performance.
 
iampoor1 said:
If you want to try a transformer (for the hecking heck of it  ;D) you can always try a 1:1 input transformer from Jensen...the mic splitter transformers come up on Ebay often for very little $.
Actually the noise figure in the 1510 datasheet uses Rs = 200 ohms and those mic splitter transformers have 3 windings so if you wired two in series for a 2:1 step-down (using the "center tap" for phantom), the lower Rs of the secondary might equate to lower noise with the 1510 compared to what you loose from the step-down. And those JT-MB splitter transfomers are pretty common and relatively inexpensive.
 
PRR said:
The WHOLE point of those SSI and THAT chips is to NOT use a costly heavy and narrowband transformer!
Pfff.

130dB CMRR at 50Hz, 3dB down at 5Hz and 80KHz, Faraday shield kills RFI stone dead, bullet proof against 48V phantom (no expensive preventative maintenance every few years to replace those leaky electrolytics at the input), galvanically isolated input (no hum loops). Work out the total cost of ownership for an electronic mic input over its life and I'll pay the 40 bucks up front for a transformer every time.

Cheers

Ian
 
If I was going to use an input transformer, I'd likely go with the same OEP's that the Gyraf G9 uses (for those of us who can't crap enough money to afford Lundahls!).

I'm going to start scratching out something based on either/or.
 
So, I slapped this together based on the design notes from THAT.

For now, just going with capacitor input coupling.  Still trying to decide where to put the pad switch (if I bother with one), the HPF,  and the polarity switch (right before the output jack, maybe?).

The first stage is set for about 54dB of gain, and the output stage is supposed to provide another 6dB.  I'm not entirely sure about the value of VR1.  I just went with what I used for a monitor controller I built recently.

THAT suggested a network of diodes on the input stage and on the output stage for IC protection.  I've seen similar things done in other design, and I'm wondering about the pros and cons of any of them.  Still researching that...

I give you....

"THAT Thing!"

 

Attachments

  • That Thing.jpg
    That Thing.jpg
    194.2 KB · Views: 49
Why do you want to control volume after the amplifier instead of changing the gain? You will easily run into headroom problems this way. Even if you want to build this for a very specific fixed purpose, the 1646 has an input impedance of only 5k and should be driven by "zero" impedance to achieve its advertised performance. So you would have to buffer the volume pot by another opamp.

Concerning the input capacitors and protection diodes: you should read the "Phantom Menace" articles.
 
I agree with volker,  also C5 is much too small and will roll off low end.

For the input stage you need to consider levels,  a loud condenser might put out 1V. At 54 dB gain that will be way distorted.
 
OK.  A couple revisions....

Variable gain on the input of the 1512.  Added a -15dB pad.  (I *think* my math is right...), and some circuit protection that THAT recommends.  Also added a polarity switch.  The only other thing I wanted to add was a HPF for a rolloff around 80-100Hz.  Just not sure where to put it.
 

Attachments

  • That Thing rev2.jpg
    That Thing rev2.jpg
    232.5 KB · Views: 34
Much better.

But you really don't need the pad. Aside from the fact that the 151X has good low noise performance, what's the point of the pad? If the input is line level and the output is line level, you have a buffer with a polarity switch.
 
squarewave said:
Much better.

But you really don't need the pad. Aside from the fact that the 151X has good low noise performance, what's the point of the pad? If the input is line level and the output is line level, you have a buffer with a polarity switch.

Excellent point, now that I think of it.  If the first stage had a fixed gain, a pad might be handy; but it's continually variable from 0-54dB, so if the input device is too hot, just turn it down!

I would imagine that to overload the input, the signal would have to exceed the supply rails of the chip, right?
 
Back
Top