Topic: DIY preamps

[SparkE!]

Op-amp notes I have:


BTW, the stock UA-5 is full of NE5532's. Most people think it's not the greatest sounding op-amp chip for microphone pre-amplifiers.

http://www.oade.com/Tapers_Section/Forum/dcboard.php?az=show_mesg&forum=100&topic_id=104&mesg_id=104&listing_type=search

The UA-5 does not use NE5532s it uses the NEC 4570 pos. There are so many great op amps out there now that if you want your own sound for a pre do a lot of research and comparison.

I'm going to try to trace the UA5 when I get a chance.  I've got some OPA2134 (TI samples!).  Either that, or I'll use AD712/AD8620 that Oade recommends.  I just have to find out what ones to replace.  Perhaps just one or two on the mic pre inputs.  Has anyone dissected one of Oade's mods?

For now I'm going to build a homemade pre based on INA217 or INA103 (Thanks, L'il Kim!!).  I'll post a "chalkboard" if I get anything working...

  Richard

Richard, when I've gone spec shopping, I've found that some of the Linear Tech op amps look like the best bang-for-the-buck.  I have not done any listening comparisons, but when I look at the OPA2134 ($2.36 at Digi-Key) at 8nV/rtHz, the AD712 ($5.76 at Digi-Key) at 16 nV/rtHz and the AD8620 ($13.33 at Digi-Key) at 6 nV/rtHz, I start thinking that the LT6231 ($3.88 at Digi-Key) at 1.1 nV/rtHz and the LT6201 ($7.38) at 0.95 nV/rtHz both look pretty damned good.  Both the LT6231 and the LT6201 have rail-to-rail inputs and outputs and both would be quite happy with the 5V supply in the UA-5.  I mean you're talking about less than 1/5th the noise of any of the op amps you've mentioned and both of them have gain/bandwidth products of over 165 MHz, which is more than 6 times that of any of the op amps you mentioned and they have the same pinout.  The prices I quoted above are for the SOIC-8 package which is the package style you'd need for the UA-5 if I remember correctly.

I'm planning to do some mods to all of my UA-5's and when I do, of course I'll tell about it somewhere here at TS.com.

I'm also interested in the parts that are actually used in the Oade mods too, but I've read that they burnish or buff the letters off the op amps so that you can't tell what they are.

The thing about the Oade mods that I reallly don't get is why they end up disabling the 1/4" inputs.  Does anyone here have any idea of why they had to do that?  It doesn't make sense to me that it should have to be that way.

[poorlyconditioned]

Op-amp notes I have:


BTW, the stock UA-5 is full of NE5532's. Most people think it's not the greatest sounding op-amp chip for microphone pre-amplifiers.

http://www.oade.com/Tapers_Section/Forum/dcboard.php?az=show_mesg&forum=100&topic_id=104&mesg_id=104&listing_type=search

The UA-5 does not use NE5532s it uses the NEC 4570 pos. There are so many great op amps out there now that if you want your own sound for a pre do a lot of research and comparison.

I'm going to try to trace the UA5 when I get a chance.  I've got some OPA2134 (TI samples!).  Either that, or I'll use AD712/AD8620 that Oade recommends.  I just have to find out what ones to replace.  Perhaps just one or two on the mic pre inputs.  Has anyone dissected one of Oade's mods?

For now I'm going to build a homemade pre based on INA217 or INA103 (Thanks, L'il Kim!!).  I'll post a "chalkboard" if I get anything working...

  Richard

Richard, when I've gone spec shopping, I've found that some of the Linear Tech op amps look like the best bang-for-the-buck.  I have not done any listening comparisons, but when I look at the OPA2134 ($2.36 at Digi-Key) at 8nV/rtHz, the AD712 ($5.76 at Digi-Key) at 16 nV/rtHz and the AD8620 ($13.33 at Digi-Key) at 6 nV/rtHz, I start thinking that the LT6231 ($3.88 at Digi-Key) at 1.1 nV/rtHz and the LT6201 ($7.38) at 0.95 nV/rtHz both look pretty damned good.  Both the LT6231 and the LT6201 have rail-to-rail inputs and outputs and both would be quite happy with the 5V supply in the UA-5.  I mean you're talking about less than 1/5th the noise of any of the op amps you've mentioned and both of them have gain/bandwidth products of over 165 MHz, which is more than 6 times that of any of the op amps you mentioned and they have the same pinout.  The prices I quoted above are for the SOIC-8 package which is the package style you'd need for the UA-5 if I remember correctly.

I'm planning to do some mods to all of my UA-5's and when I do, of course I'll tell about it somewhere here at TS.com.

I'm also interested in the parts that are actually used in the Oade mods too, but I've read that they burnish or buff the letters off the op amps so that you can't tell what they are.

The thing about the Oade mods that I reallly don't get is why they end up disabling the 1/4" inputs.  Does anyone here have any idea of why they had to do that?  It doesn't make sense to me that it should have to be that way.

Thanks (and +T) for the tips.  I'd be grateful for any help, both on opamp selection and tracing the UA5 input circuitry.  I believe it runs on something more than 5V, maybe +/-9V.  There are also some transistors, so the mics may or may not go directly into an opamp.  I can't disassemble mine right now (I rely on it, and I've covered/locked a lot of the connectors with hot melt glue!), but I may be getting a second one soon.  If so, I'll certainly be digging in...

  Richard

[Nick's Picks]

I recently picked up one of these on eBay..for the sole purpose of learning how to MOD a preamp.
http://www.musiciansfriend.com/srs7/src=SearchDex/search/detail/base_pid/180330
only, i got mine dirt cheap and used.

[SparkE!]

Thanks (and +T) for the tips.  I'd be grateful for any help, both on opamp selection and tracing the UA5 input circuitry.  I believe it runs on something more than 5V, maybe +/-9V.  There are also some transistors, so the mics may or may not go directly into an opamp.  I can't disassemble mine right now (I rely on it, and I've covered/locked a lot of the connectors with hot melt glue!), but I may be getting a second one soon.  If so, I'll certainly be digging in...

  Richard

As I recall, the analog audio path is powered by +/- 5V, but I might be remembering wrong.  I know that there's a switching power supply module, all potted in black epoxy that has a +5V output, but I don't remember if it's also responsible for the -5V.

Also, there used to ve a partial input schematic of one channel on the Oade boards, but it does not seem to be there anymore

[Todd R]

Obviously specs don't tell you everything.  But in addition to noise, I'd consider speed (slew rate).  That will give you some kind of indication of the amp's ability to keep up with the changing dynamics of the music and help create a solid, 3-dimensional soundstage.

As to the Oade mods, I seem to remember him posting a long while back that the mic in needed to be disabled for power reasons.  I think the op amps he was using drew a lot more current than the stock ones, so rather than rebuilding the DC converter circuit to provide more current capacity, he disabled the mic-in section to save on power.

Which brings up the idea that in addition to noise, slew rate, and rail-voltage reqts;, you should also consider current draw when selecting op amps.  For our purposes with field recording, it sure would be nice to use op amps that run on 0v and +5v (or even +3v) and have minimal current draw.  That does narrow the field quite a bit though.

[SparkE!]

Todd, the LT6231 has a 70V/us (typical) slew rate and the LT6201 has 50V/us (typical) slew rate with +/-5V power supply, as compared with the UA-5's stock NEC uPC4570 with only a 7V/us (typical) slew rate, so the Linear Tech op amps that I mentioned are plenty fast enough.

And it's not the mic input that's disabled.  It's the 1/4" input which uses exactly the same amps.  (See the schematic below on the inputs labeled as "TRS".  The regular XLR inputs come through 2 caps to protect the TRS inputs from the phantom power path.)

Here's the UA-5 left channel input schematic from the Oade boards:



If it doesn't show up for you, go here:

http://64.27.110.248/Tapers_Section/images/UA-5leftpreaudiopath17jan03.jpg

[poorlyconditioned]

Todd, the LT6231 has a 70V/us (typical) slew rate and the LT6201 has 50V/us (typical) slew rate with +/-5V power supply, as compared with the UA-5's stock NEC uPC4570 with only a 7V/us (typical) slew rate, so the Linear Tech op amps that I mentioned are plenty fast enough.

And it's not the mic input that's disabled.  It's the 1/4" input which uses exactly the same amps.  (See the schematic below on the inputs labeled as "TRS".  The regular XLR inputs come through 2 caps to protect the TRS inputs from the phantom power path.)

Here's the UA-5 left channel input schematic from the Oade boards:



If it doesn't show up for you, go here:

http://64.27.110.248/Tapers_Section/images/UA-5leftpreaudiopath17jan03.jpg

Thanks (and another +T).  Exactly what I wanted to see.  So, it looks like a standard "intrumentation amp" layout (two sides done separately, then a single amp on the output).  It also looks easy to hack.  Another thought is to replace the first few amps by a single INA217.  I believe that will run off of +/-5V too.  Anyway, I'm going to try the LT chips first.

Question: Do the TRS inputs go through a separate or the same circuit?  I'm curious, because as they're shown now they would have lots (maybe too much?) gain, right?

  Richard

[Chuck]

One thing I see, is that if you disable the TRS inputs, you can do away with C80 and C77. Eliminating any capacitors in the audio path is a good thing.

[SparkE!]

One thing I see, is that if you disable the TRS inputs, you can do away with C80 and C77. Eliminating any capacitors in the audio path is a good thing.

Yeah, that allows you to use capacitors C70 and C85 whose capacitance is 1/2 that of the original.  When you're dealing with plastic film capacitors, physical size is going to be an issue.  So, if you can use 1/2 as many capacitors and the capacitors that you use have values that are 1/2 that of the ones they replace, that's going to help avoid running into space restrictions.

[Todd R]

Thanks for all the info, SparkE!

Yep, and cancel everything I said about the Oade mods -- somehow I was thinking we were talking about the modSBM1, not the UA5.

[SparkE!]

Something that's not shown on the schematic above is the Hi-Z/Lo-Z switch.  So there are some parts near the inputs that are missing on the schematic.

Regardless, the UA-5's abbreviated input schematic, shown above, is typical of a balanced input pre-amp.  You could use that topology as the basis for a DIY pre-amp, but use lower noise op amps and coupling caps.  You might even consider using metal film resistors in the feedback paths for the input amps to help reduce the  1/f noise that you'd otherwise get from typical carbon film resistors.  It might also be a good idea to reduce the input impedance by intentionally loading the inputs with resistors to ground.  Remember that thermal noise voltage in a resistor is (4kTBR)^1/2 where k is Boltzmann's constant (1.38*10^-23 Joules/degree Kelvin), T is the temperature in degrees Kelvin (273 + degrees C), B is the signal bandwidth and R is the resistance of the resistor in Ohms.  Unfortunately, the lower you make R, the higher value you must use for the input coupling capacitor because the lower cutoff frequency of the input is 1/(2*pi*RC) where C is the capacitance in farads and R is the resistance in Ohms.

[poorlyconditioned]

Something that's not shown on the schematic above is the Hi-Z/Lo-Z switch.  So there are some parts near the inputs that are missing on the schematic.

Regardless, the UA-5's abbreviated input schematic, shown above, is typical of a balanced input pre-amp.  You could use that topology as the basis for a DIY pre-amp, but use lower noise op amps and coupling caps.  You might even consider using metal film resistors in the feedback paths for the input amps to help reduce the  1/f noise that you'd otherwise get from typical carbon film resistors.  It might also be a good idea to reduce the input impedance by intentionally loading the inputs with resistors to ground.  Remember that thermal noise voltage in a resistor is (4kTBR)^1/2 where k is Boltzmann's constant (1.38*10^-23 Joules/degree Kelvin), T is the temperature in degrees Kelvin (273 + degrees C), B is the signal bandwidth and R is the resistance of the resistor in Ohms.  Unfortunately, the lower you make R, the higher value you must use for the input coupling capacitor because the lower cutoff frequency of the input is 1/(2*pi*RC) where C is the capacitance in farads and R is the resistance in Ohms.

Excellent thread.  Let's keep the DIY going...

I agree about the load resistors.  They may be there in the UA5.  I'll look at this if/when I trace the schematic.

As far as a DIY preamp, I think there are two good choices.  One is to use a *single* low-power opamp, like the LT6234.  This is similar to the other LT series, except 1.1mA for each amp, which is great for long battery life.  I suggest a single amp because likely were running an unbalanced signal in.

The other option is to use an INA217 (or the better spec'ed INA103) which is a complete, on-chip instrumentation amp (three opamps above, and it requires only a single external gain resistor).  This requires a split supply, at least +/-5V, and about 10mA per channel.  I'm thinking a single 9V alkaline battery for the first option, and a Walmart battery (7 to 9V) plus voltage multiplier/inverter (LT1054) for the second option.

I'll report back here (with chalkboard schematics) if I get anything working...

  Richard

[Chuck]

I have a simple mic-pre I built around an op-amp in a difference circuit, using (2) 9v batteries to optain the split voltage.
I recently have been reading about headphone pre-amps built using virtual ground circuits using the TLE2426. These circuits can split a DC voltage in to two equal and importantly, stable +/- voltages. Of course there are also single supply op-amps out that don't need this treatment. The headphone amp guys have built a little cottage industry around perfecting, then making the circuit boards and helping people built there own headphone amps.

I've been reading a lot here lately:

http://tangentsoft.net/elec/

Chuck

[SparkE!]

For virtual ground circuits, I've always used something like the op amp-based virtual ground driver they show over there at tangentsoft, except that I put a capacitor between the positive input and ground.  Here's their circuit:



If there are going to be large, impulsive ground currents running, then I'll add a buffer stage with a unity gain buffer driving the virtual ground node.

It seems to me that the main advantage of those "rail splitter" ICs is that they are tiny.  I can live with 10 or 20 mV of offset in the virtual ground.  The outputs of your op amps are not going to be symmetric to that degree anyway.  They go high impedance when they near the rail and the impedance change is not symmetric.  If you want to truly operate rail to rail, it will limit the amount of gain that you can get per stage or you'll trade gain for harmonic distortion.

[Chuck]

Yeah, I started reading about DC offset when using the (2) 9v batteries and measured up to .2 v offset due to different battery voltages!

I'd also like to try some of the op-amps those Head-Fi guys are using. There are lots of new op-amps out since I started doing this about 15 years ago.

Maybe we could come up with a circuit design, test it and optimize it and make some project kits or something. I know something about the circuitry, but I'm not really up on all the op-amp choices available now.

Also, I always choked on the phantom power circuit design. There was always noise introduced due to the voltage steping.
So, I really like the phantom power scheme the Oade brick pre-amps use. Straight pure DC power from batteries.
This isn't complicated, and it sounds good.