Current drive audio amp

[dr pepper]

As part of another project also on this forum I'm driving field coils with a audio amp chip, the Tda7265.
The bandwidth of the chip is good enough, but the inductance of my field coils is so high that bandwidth is an issue, peak amplitude is around 4V, and the supply is 12V, maybe a little higher if required.
So I was thinking of configuring the amp as current drive, or mixed mode feedback, I could use an rc to equalize the coils impedance to keep the amp stable.
I dont think I've messed with current drive amps before, any suggestions on circuits?

 

[ronsimpson]

Years ago I wrote a paper on current amplifiers driving inductive coils. (vertical deflection amplifiers)
I can't find the paper now. I will look more.
In a vertical deflection amp I often used a 1 ohm (more or less) resistor on the cold end of the coil to measure current. That is where the feed back comes from. You want current not voltage in the feed back.

If you give me the current (p-p or pk or rms), coil resistance, coil inductance I can work something up. Also I want to know what type of signal you want on the input. square/saw tooth/etc and signal size.
---edited----
Do you need accuracy! or will a little temperature drift be OK. (wire temperature of the coil)

 

[ronsimpson]

I just picked a amp. Not like yours. R4=wire resistance. R2 to measure current.
R1/R2 points at 20db gain .
BUT the amp sees R3 and R4 so it must make 30db of gain at DC.
L1 "opens up" at high high frequency. So the gain starts out at 30db at 0 to 10hz. 45db at 100hz 65 at 1khz 65db at 10k.
So the higher the frequency the more voltage gain is needed to get a signal through L1.
Now place the amp's gain/frequency curve on top of this. Its gain is very high at DC and at some point starts down and hits zero at some high point.
When the "gain we need" and the "max gain of the amp" cross the circuit will be unstable, at that frequency. That is the point we need to stay away from.
R5 feeds the "voltage" back to the amp. It limit the max gain. I have used R5 like this because the coils I drive have winding capacitance and that adds more problems. (R5 dampens the winding cap) I have not tried it but I think the resistor can go from OUT to -IN and get the same frequency gain fix.

What I do is build a real circuit with out R5. Frequency sweep the circuit and watch it oscillate. OR Step it and watch it ring. By knowing where it wants to "ring" I then now what to do with R5.

 

[dr pepper]

I wasnt expecting someone with that kind of knowledge.
Ok I'll find out that info as soon as I get back to the 'shop.
I'd like to use this for vector graphics, though it may not be practical.
Accuracy isnt all that important this isnt an oscilloscope.
I've done this before with magnetic deflection, only that time I rewound the yoke, less turns reduced the inductance and dc resistance making it possible to run the thing with a voltage feedback amp, this time however the yoke is one potted assembly so I dont have that option.

 

[dr pepper]

Ok I'm back.
The waveform from vector graphics is pretty random, bandwidth I'd say is around 10kc.
There are 2 coils:
X: 34mH / 16.6 ohms / just under 1mA full deflection / around 2.5v Rms
Y: 60mH / 55.0 ohms / just under 1mA full deflection / around 2.5v Rms
They seem like very high values to push 10kc through, sounds like its all about the power supply voltage.
Thanks for your help Ron.

 

[dr pepper]

After an evenings lab session I didnt get good results.
The 34mH coil responds well, tuning the circuit got flat response to around 10kc, however the 60mH coil does not respond well, the best I could get was only around 3kc, the image using an arduino generating vector graphics is pants.
So reality check it looks as though this tube cannot do what I want, its tuned for slow scan radar, and thats why it has long persistence phosphor.
However there is one more thing I thought of and might try, I have a few salvaged yokes from Tv's, one looks as though it'll fit the neck of the Crt, last time I built a vector monitor I used a standard Tv tube and to speed up the vertical which in a Tv only runs at 25hz I rewound it with much less turns, massively speeding it up and making the tube respond well.
So I think I'll chuck a yoke on this tube from a Tv and see if I can get anything like reasonable linearity, the coil form looks slightly different as a 14" tube has a higher deflection angle, however what'll do a lot might well do a little.

 

[ronsimpson]

X: 34mH / 16.6 ohms / just under 1mA full deflection / around 2.5v Rms
Y: 60mH / 55.0 ohms / just under 1mA full deflection / around 2.5v Rms

I am surprised that the 60mH coil needs 1mA. I think the turns are larger so the current should be down. Looks like the wire size is smaller for the 60mH coil.

If you can drive 34mH to 1mA at 10khz, then it should take twice the supply voltage to drive 60mH to the same 1mA at 10khz.
If both amplifiers have the same supply voltage then; 34mH=10khz, 60mH=5khz.

It could be that the two coils are wound in different modes.
1)Below you can see the wire is wound around the core. " Toroidal "

2) This coil is not wound around the core. "saddle saddle"

 

[dr pepper]

My old sig gen could well be out, the 34mH coil dips at 9kc, the 60mH coil could well be 4.5kc I didnt bother being that precise.
I approximated the current consumption from the input power on my power supply.
The yoke potting goop is semi transparent, I'm pretty sure theres no core, looks like the coils were wound on a former then caked in epoxy, obviously the two windings are different, hard to say if they are similar to the yoke you posted, the one I thinking of trying looks just like that one, when I used that kind on my last vector monitor I rewound the vertical to get it so's I could increase the frequency response.
Yes I'd need twice as much voltage to drive the 60mH, I did that much messing about I cant remember what I got measurements wise, it could be that the amp is clipping on the 60mH coil before its gets any higher than 3 - 4.5kc, or maybe its limited by slew rate or Bwp, certainly at 1ma it doesnt need to be a power amp, a Lf358 works Ok.
I allready put together a smps for the Crt, wished I hadnt now, might need another couple of rails for the deflection amp.

 

[dr pepper]

I fiddled around for a while with the sim, I changed the i/p resistor to 100k for a gain of 0.1, then replaced the current sense resistor with a 470r.
I now get 1ma through the coil with an i/p voltage of 5v, as you can see on the chart which goes from 10hz to 15khz the current through the coil only changes by 20ua over that range, drive voltage at 10hz is 0.5v and 15kc its 6v, jst above 15kc theres a goofy resonance thing going on, fiddling with the values got that high enough to be outa range, the circuit might need a low pass to make sure it doesnt do something funky (unless damping it works).
The sim shows the coil will ring with its 200pf of stray capacitance, however a 6k2 and a 2u2 cap snubber damps that down, seems like it ought to work at least ball park.

Seems that with partsim you cant really monitor current by using a voltage across a resistor, you have to use a current probe.

 

[dr pepper]

Ok having tried the circuit in practice it works well, nice & linear, but my current measurements were out, way out, it takes more like 30ma for full deflection.
And to get full deflection at that current I'd need 100v at 10kc, bummer.
I had an idea of using a 100v line audio transformer, but then there would be issues with compensation, so maybe its time to try the modded tv yoke., theres a yellow filter with range marker lines over the Crt so I could if I really wanted chuck the Crt out the radar set, take the one out of the 5" Crt b&w Tv I used for initial testing & bung that in behind the yellow filter, probably be a lot easier.

 

[ronsimpson]

10kc, bummer.

Are you looking at sweeping like a TV or more like a radar scope?
Radar CRTs have a persistence of many seconds. I remember one that we redrew every 9 seconds. (much slow video rate)

 

[dr pepper]

Neither, vector graphics are very much like drawing with a pen & pencil.
yes persistence is 1 second on this Crt, after glow is a slightly different colour.

 

[ronsimpson]

Drawing a "radar" picture took little power.
Going back and drawing text on top took much power.

 

[dr pepper]

Yes most likely.
Text on screen wasnt heard of when this set was made, hence my trouble.