[sdiy] Gilbert Cells and VCAs with Predistortion

[Ryan Williams]

Magnus Danielson wrote:
> Build some traditional OTA/VCA designs onto the same board and make the
> comparision in the same project.

thats a good idea. I'll do it.

> A classic way to do THD is to subtract the output signal with the input signal
> and balance them towards minimum. The residue signal should be the non-harmonic
> (not all true, but ah well, it's an approximation) The ratio between the
> signals is the THD.
> 
> On my network-analyzer/spec I have a little program that measure the 10
> overtones and calculates the THD from that. Lovely to have gear that can be
> taught to think for themselfs... ;O)

The network analyzer's at school are from Agilent.  I'm not sure if the 
one they have measures low frequencies. I'll have to get the manual for 
it and find out. The spectrum analayzers that I have seen are similar. 
I know they measure DC to 1GHz but what I'm not sure about is the input 
level. I think these tools are meant for RF.

They do have some real nice Agilent oscilloscopes. I think I could 
achieve this by recording data from the scope to a computer and then 
running the FFT in something like matlab?

One other option I thought about yesterday was national instruments 
labview. The school has several of these systems. I found a tutorial on 
using it for audio performance analysis, but I'll have to read about it 
and find out how precice it can measure.

If you or anyone else is hoping to see my results, I must warn you that 
it will probably be a while. I have several other synth, and school 
projects going on at the moment. but, I really want to do this as soon 
as I can.

>>I still have about 15 THAT2180 VCAs but I'm saving them for situations 
>>where I want exponential control.  I'll probably just use them in other 
>>synth modules that need an on board VCA.
> 
> 
> Well, the VCA is 2 PNP and 2 NPN and an inverter as "active" components, so it
> was easy enought to do. The linear control input can be acheived by using a
> diode to logarithmize the signal. Remember, over there some temperature
> dependence might not be the worst thing in the world.

I suppose your are right. I probably would not even know the difference 
if there was some temperature dependence. I did not really intend to 
make a VCA circuit when this started. I was studying the gilbert cell to 
learn about 4-quadrant multipliers, not OTAs and VCAs.  I guess my 
excuse about nonlinear control of the THAT chips is no good.  For me 
this is just an educational exersize to improve my knowledge and 
understanding of nonlinear circuits, but it would be nice to come up 
with an alternative to the OTAs that are slowly dying out.

-ryan williams