Help with testing inductance in a DC circuit

[edcorusa]

We are presently manufacturing various transformers that are gapped to allow DC currents to flow. We use the Hanna curve and have developed a computer program that allows for the gapping, core size, etc. We are having a problem with testing of the inductance with different levels of DC flowing through them. We very good test equipment for all other tests and a precision DC supply. Does anyone know of good test equipment that will test inductance with DC flowing in the circuit?

 

[burdij]

Could you couple your AC inductance bridge to the inductor with capacitors and put a large choke in series with the DC current supply to block the AC from the bridge from flowing through the relatively low impedance of the power supply?

 

[edanderson]

why not measure the inductance of another winding? flow the current through the primary and measure the inductance on the secondary, for example. since you know the turns ratios of all the windings, you should be able to infer the inductance of one winding from another? maybe i don't understand some detail that would limit this approach's usefulness...

ed

 

[burdij]

Here is a method for measuring inductance in a reactor carrying a large current. It requires a gapped coupling transformer that can handle the DC current without saturating while passing the test AC waveform. This transformer shouldn't be too hard to build for smaller current levels at a typical testing frequency of 1KHz.

http://www.edn.com/archives/1995/070695/14di4.htm

 

[mikep]

we recently had an agilent impedance meter on demo that could do this kind of measurement directly (the instrument supplies the bias). I dont remember the model number but it was something like this:

http://www.metrictest.com/product_info.jsp?mfgmdl=HP%204263B

very cool if you can justify the expense. in the end we couldn't

 

[PRR]

> various transformers that are gapped to allow DC currents to flow

For vacuum-tube plate circuits. For the Usual Tubes.

(Could be transistor, but the sexy market is tubular.)

Rig a 6L6, a 250V-500V supply, an adjustable cathode bias resistance (unbypassed), and a high-level signal generator.

Load primary or secondary with nominal load.

In most conditions, 6L6 plate resistance will be almost 10 times higher than nominal load. With unbypassed cathode resistance, more than 10 times. You can ballpark this by changing the load with constant grid drive. You could pretty much ignore it, or apply a 10% fudge.

Bring 6L6 up to transformer's nominal current.

Bring audio grid drive up to a fraction of a Watt.

Sweep frequency down from midband to the -3dB point. F and R and a couple Pi give effective L.

"R" would be dummy load R in parallel with 6L6 effective plate resistance. But this error will be OTOO 10%. I hope you take your prototype measurements and deduct ~~20% for production leeway and protection against too-fussy customers. Measure 10H, promise 8H, deliver 9.0-9.5H.

Try with part-Watt, also at "full power", and at some intermediate signal voltage. Watch waveshape change with frequency and power. No-DC transformers will respond different for small or large signal. Big-DC transformers may not. (Nobody is paying me to find out.)

Nudge 6L6 DC current a bit high of Rated Current and spot-check. If current 20% high gives inductance 20% low, no big deal. If the inductance "falls apart" for typical tube tolerances, you need more robust design (less sharp pencil, more iron).

Customers using pentodes with bypassed cathodes will get response a hair better than L and load say.

Customers using triodes at typical loading will get response nearly 3 times better than L and load compute to (triode Rp is usually less than half of load).

An alternative is to use a big MOSFET in a tube-like circuit. Rp is infinite for all practical purpose, and no heater supply. You do have to run high DC voltage to verify low distortion at high signal levels. You may kill a few MOSFETs because while they can be bought for 1,000V, glitch-kicks on 250V or 500V DC supply in a low-loss choke will flyback couple-KV spikes.

There are more precision ways to measure L in presence of DC. Most require a choke MUCH better than the one you are testing. And if your market is tubework, or understands that this iron is sold for such work, the in-circuit 6L6 measurements are what they need to know.

When you get bored with 6L6, I have a scheme for 813 at 45 Watts output single-ended. Guitar-band... full bass would be too much for my floor. I suspect even then the weight and cost are beyond my tolerance. But fun to think about while tuning my puny 16W SE 6550.

Put 4/8/16 secondaries on your iron. I know why that is a "poor idea". But often I can't know what speaker I will have to drive until I get to the gig. And saves you from stocking three SKUs for every basic design.