Simple unbalancing and balancing line amp

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
And if you are going to do that, might as well get one with buffers so that the CMRR is better, like INA1650 or INA1651.
Hard to have more simplicity than one device.
Wow ! I had never heard of that chip, it looks like a gift from heaven ! That's exactly what I need in a single component package ! And it's even stereo. Thanks !

For the input, yes, why not, but I wouldn't recommend that for output.
For the output I meant a pair of 1:1 Edcor right after that single 5532. It seems simple, space isn't a issue, but I don't want to use 4 transformers on such "low end" device. This sounds like a good in between solution.
I just need to figure out how to make gain changes with a pot in that circuit.

Feels like I'm getting close, thanks !
 
The 2520 is not like any op-amp, though I remember someone from Signetics saying that they tried to make a chip out of it (5534) but it wasn't stable. Many op-amps say they are the same but aren't. Their is one that is, and it's spelled backwards... It's a high gain design that if modeled on spice is latched to the negative rail, in real life it always has negative offset... The slew rate (in several configurations) is only 2.5V/usec, but with a 1:2 transformer, that doubles...
 
I went for simplicity : I took out my "Small Signal Audio Design" book by Douglas Self, and got interested in these : what's your opinion ?


Input :

View attachment 97341


Output, 2 options :
View attachment 97340View attachment 97342
In both option, I'll see how I can make gain variation, i'll study the 5532 datasheet. 2nd option is almost too simple to be true, it def has my vote for the moment :)
I used to use servo DC offset amps in the API stuff, they aren't worth it unless you have a lot of offset. The design will actually put DC on the input if you don't have caps, so you might as well have a cap on the output. They do seem to effect transients sometimes. The slower you make them, the longer the console swings back and fourth until they settle.

I use the THAT 1256 on the input, which sounds really nice, and 2) 1206 with +/- reversed on the second one for balanced outputs. Yes, if you short one side you lose half, but they sound better than the 1646 output thing, which I've never liked.

No record ever failed because someone picked the wrong input stage...
 
Hey Paul, thanks for chiming in !

I am confused again now. Are you saying the THAT 1256 is preferable to the TI INA1650 for the input ?
I see the 1256 has -6dB gain, I assume this is to compensate for the unbalancing of the signal, to keep an overall unity gain "feeling" ?

Also, if I understand correctly, you recommend that I use a THAT 1206 for the output. I've read the datasheet and I feel like I won't be able to play with the gain of the chip and have, say, 20dB gain to compensante for the my unit's gain reduction. If my assumption is correct, then I should be stick with the Douglas Self design using the 5532 + transformer, right ?

No record ever failed because someone picked the wrong input stage...
Haha, true !
 
I used to use servo DC offset amps in the API stuff, they aren't worth it unless you have a lot of offset. The design will actually put DC on the input if you don't have caps, so you might as well have a cap on the output.
the benefit of servo amps is the ability to use well behaved small film caps , in place of less well behaved large electrolytics.
They do seem to effect transients sometimes.
Not if properly executed. A good servo topology involves a passive LPF, so extreme HF is scrubbed harmlessly.
The slower you make them, the longer the console swings back and fourth until they settle.
not if they are well damped and stable..
I use the THAT 1256 on the input, which sounds really nice, and 2) 1206 with +/- reversed on the second one for balanced outputs. Yes, if you short one side you lose half, but they sound better than the 1646 output thing, which I've never liked.

No record ever failed because someone picked the wrong input stage...
Indeed....

JR
 
Are you saying the THAT 1256 is preferable to the TI INA1650 for the input ?

That 1256 is just a diff-amp with trimmed resistors, same as INA134. Any kind of diff-amp based input is going to be very sensitive to any imbalances on the drive side impedance. The INA1650 will have better CMRR with slightly unbalanced drive. The That InGenius parts (120x family) will have even better CMRR robustness, at the cost of a couple of extra components, maybe slightly higher price.

I'm fine with soldering some SMD, but OK

Don't let the fact that Abbey's eyes are aging out of being able to work on TSSOP dissuade you. ;)
I have to use strong magnifiers myself, but can still solder them for the time being.

I see the 1256 has -6dB gain

Most of the parts come in variants with 0dB, -3dB, or -6dB. Pick the one that works for your design based on the gain structure that you need.

I feel like I won't be able to play with the gain of the chip and have, say, 20dB gain to compensante for the my unit's gain reduction

For any of the integrated balanced line drivers you will need an additional op-amp stage if you want adjustable gain. You should have an additional op-amp as a buffer anyway if your input signal is not being driven directly by an op-amp output. It isn't clear if you are going to tap into the internal circuitry, or just connect this to the existing output signal. If the latter, outputs usually have a build out resistor, so you would need to buffer that before driving any of the integrated differential output devices.

Of course just using impedance balanced output with asymmetric drive is a valid choice as well. That hasn't really come up, but has been discussed plenty of times on other threads. With 12V power supplies you should be able to get to about +18 dBu to almost +20 dBu depending on the output amps you use. You never really defined your level requirements, so not clear whether that is enough for you or not.
 
Thanks. I don’t want to modify the DBX118 too much, so I’ll just use its stocks RCA input and output.
Before the unit’s input : I want a simple unbalancing stage with 0dB gain.
After the unit’s output : I want to be able to add gain from 0dB to about 20dB, and then balancing. +18dBu is plenty enough
 
I'm fine with soldering some SMD, but OK. Do you suggest I go with the THAT 1256 Paul is talking about ?
The 125x series is a downgraded version of the 124x. I've never used the 125x, always the 124x. The only significant difference is the nominal CMRR is 50dB instead of 90. It may seem a lot, but in a studio environment, typically 40dB is enough.
 
Last edited:
Hello everyone,

I've done some homework, and here is what I came up with. This is pretty simple, I use the THAT 1250 for unbalacing, then go to the DBX118, then out of it into Douglas Self's baxandall active volume control "Precision Preamplifier 96", and then into his transformer output module. Max gain would be 17dB, which is perfect for me. My question is : can I skip the 2nd 5532 and directly use the inverting amp (half of U1) to drive the output transformer (Edcor 1:1) ?
I am drawing this in Eagle, working on the PCB and then out to Fusion 360 to cut if with a milling machine.

Thanks

Thomas

ps:
1) sorry for the DIP8 view in the schematic instead of two halves of the 5532, I could find a decent 5532 in the librairy so I had to use a generic DIP8 socket
2) didn't pay much attention to phase when hooking up to the in & out connectors, it's more about the circuit concept and component values here (straight out of Self's book)
 

Attachments

  • 118_IO_v1.jpeg
    118_IO_v1.jpeg
    152.2 KB · Views: 4
My question is : can I skip the 2nd 5532 and directly use the inverting amp (half of U1) to drive the output transformer (Edcor 1:1) ?
Yes. Actually the performance may be slightly inferior when the opamp operates at maximum gain and drives the xfmr, but the reduction of number of stages may very well compensate it.

I actually question the connection to the transformer via a 47r resistor.
At LF, the xfmr wants to see the smallest possible impedance, so a direct connection would be preferrable.
However, there's the risk of the cumulated offset voltages resulting in significant DC current in the primary, which would shift the magnetizing curve and generate distortion (2nd order).
Using a capacitor would solve the DC issue, but would result in a VLF response hump, that may cause intermodulation in the subsequent equipment. The usual fix is to use a huge capacitor, that would shift the hump under 3-4Hz.
Then there is the issue of stability, because the xfmr presents a significant shunt capacitance, which can be solved with inserting a ferrite bead.

The 47r is a cheap ans dirty fix for most of these issues except the VLF distortion.

Short of using a servo to attenuate offsets, a quite valid solution is to install a large capacitor in parallels with the 47r resistor.
 
Last edited:
Thanks Abbey
I'm impressed by your knowledge and the clarity of your explanations :)
What's a large capacitor in that position ? I mean: what value ?
I'm back to Eagle, now :)

also, 2nd order harmonic distortion, it depends how much we are talking about, but I sounds kind of desirable haha. And the DBX118 is already quite a dirty box, at least I like it better when it's distorted.
 
What's a large capacitor in that position ? I mean: what value ?
It depends on the inductance of the xfmr's primary. The resonant frequency is given by F=1/2pi. sqrtLC
A xfmr rated at 600 ohms should have at least 5H of inductance.
With a 220uF capacitor, the resonant frequency is about 5Hz. If the circuit that drives it has enough attenuation at 5Hz, there is no significant risk of intermodulation distortion in the subsequent stages.
But some mic preamps have a response that goes beyond 5Hz, so it is advisable to reject the resonance further, by increasing the capacitor value.
The amplitude of the hump can be controlled by loading the output, which may put pressure on teh output stage. A better fix is to put a resistor in parallels with the capacitor. 47r is a good starting value.
also, 2nd order harmonic distortion, it depends how much we are talking about, but I sounds kind of desirable haha. And the DBX118 is already quite a dirty box, at least I like it better when it's distorted.
It's a matter of personal taste. Colour upon colour may be too much colour...
 
Last edited:
Hello everyone,

I've done some homework, and here is what I came up with. This is pretty simple, I use the THAT 1250 for unbalacing, then go to the DBX118, then out of it into Douglas Self's baxandall active volume control "Precision Preamplifier 96", and then into his transformer output module. Max gain would be 17dB, which is perfect for me. My question is : can I skip the 2nd 5532 and directly use the inverting amp (half of U1) to drive the output transformer (Edcor 1:1) ?
I am drawing this in Eagle, working on the PCB and then out to Fusion 360 to cut if with a milling machine.

Thanks

Thomas

ps:
1) sorry for the DIP8 view in the schematic instead of two halves of the 5532, I could find a decent 5532 in the librairy so I had to use a generic DIP8 socket
2) didn't pay much attention to phase when hooking up to the in & out connectors, it's more about the circuit concept and component values here (straight out of Self's book)
You have mentioned using "2 halves" of a 5532, but were you actually meaning using 2 separate 5534's? I haven't been able to make any sense out of the schematic while comparing your schematic to the 5532 datasheet. Is it possible for you to update your schematic using either 2 5534's or showing how your schematic is "supposed to be" using the 5532? That would be most useful!!!

By the way.....are you planning on placing your PCB into a small enclosure? Or, mounting it to the inside rear-panel of a mixing console? That would determine what types of connectors you would need to specify, right? THANKS!!!

/
 
using "2 halves" of a 5532 usually means using the two opamps in the package.
THANK YOU!!! I am well aware that the 5532 has two circuits within a single package!!! I have been using them ever since they were first introduced into the market back in the early 1980s. My question stems from the fact that his schematic shows the use of the same device pins for each half of the package. In other words.....his "L_U1 Pin-1" (assuming that is actually "U1-A") is shown as being an output and connected to "L_C4", meanwhile.....his "L_U2" (assuming that is actually "U1-B") is - ALSO - showing "Pin-1" as an output feeding "L_R5". So.....how is it possible that "Pin-1" of a device that contains two separate circuits can manage to feed two separate circuits? That's why I asked about the possibility of the ICs being 5534s. If so, then what is shown would make sense!!! Or, maybe it's just me!!!

Meanwhile.....as I am looking at the TEXAS Instruments "NE5532" datasheet here in front of me, it is showing that "Pin-1" has the function of "1OUT" and that "Pin-7" has the function of being "2OUT"!!! Or, perhaps I am just not understanding the TI datasheet correctly, ya think?

thomasdf: -- Again.....I am probably completely wrong with this observation, but.....you may want to flip Pins-1 & 3 on your output connectors "BAL_L_OUT" and "BAL_R_OUT", as you are showing "Pin-3" of those connectors as being GROUND and on the XLR connectors that I have always been familiar with, GROUND is always "Pin-1" on an XLR connector. Again.....maybe it's just me!!!

/
 
Back
Top