Audio Transformer Inductance
- Thread starter ruffrecords
- Start date
Help Support GroupDIY Audio Forum:
It's not necessary to use an inductance meter for audio transformers.
Use a signal generator and meter, a suitably high resistor, and you're good to go; takes a little time to adjust and calibrate, but it gives you much more info than an LC meter that has 3 test frequencies at best and only one (unknown) signal level.
Alternatively, you may want to measure current through a suitably small resistor; that allows more power to be injected in the xfmr.
As has been already mentioned, inductance varies considerably with frequency and level. Doing measurements this way shows that clearly. For high-power output xfmrs, there is not enough power to really show these effects, though.
LC meters can be justified only for rough evaluation and production control.
Use a signal generator and meter, a suitably high resistor, and you're good to go; takes a little time to adjust and calibrate, but it gives you much more info than an LC meter that has 3 test frequencies at best and only one (unknown) signal level.
Alternatively, you may want to measure current through a suitably small resistor; that allows more power to be injected in the xfmr.
As has been already mentioned, inductance varies considerably with frequency and level. Doing measurements this way shows that clearly. For high-power output xfmrs, there is not enough power to really show these effects, though.
LC meters can be justified only for rough evaluation and production control.
abbey road d enfer said:
Well, for a hobbyist like myself, rough evaluation is a scientific result, and it really boils down to what's easier for you - set up a generator with a resistor and a meter and learn how to measure and interpret the results or spend a couple hundred on a LCR meter.
:
This meter has 5 frequencies (100/120/1k/10k/100k) and the signal level (measured) is about 0.6v RMS, good enough for a weekend audio-warrior...
As always, it's a matter of compromise and expectations.
When I was involved with graphic EQ's, I never needed an LC meter for design and prototyping, always used the AP. When production started, I bought a Racal Dana LC meter, because it was quick. The value indicated by it was so off the mark I had to spend some time creating a procedure usable by workers.
The 0.6V rms is an open voltage. What is this voltage when you measure let's say a 2H inductor? If this was constant voltage, you could not measure accurately mic xfmr primaries at the lowest frequency. That's where the most interesting things happen, how the inductance decreases when the level increases and frequency gets lower.
When I was involved with graphic EQ's, I never needed an LC meter for design and prototyping, always used the AP. When production started, I bought a Racal Dana LC meter, because it was quick. The value indicated by it was so off the mark I had to spend some time creating a procedure usable by workers.
The 0.6V rms is an open voltage. What is this voltage when you measure let's say a 2H inductor? If this was constant voltage, you could not measure accurately mic xfmr primaries at the lowest frequency. That's where the most interesting things happen, how the inductance decreases when the level increases and frequency gets lower.
I do have a high level input transformer from radio transmitter that gives a relatively paltry L reading using a handheld meter, yet when swept shows solid bottom response that doesn't match the meter's numbers very well. This particular transformer was for some reason the exception to all the other input transformers I measured and the only difference I can think of may be in level handling capability.
What is the impedance of the generator you use for your test?lassoharp said:
My question was:
What is the impedance that drives the transformer when you do a sweep (quote: "yet when swept shows solid bottom response" )
The LC meter impedance, although not specified, is high.
A transformer's LF response depends not only on the absolute value of the inductance, it is the ratio of the primary impedance to the source impedance that matters.
For 200 ohm impedance (mic input) you need about 5H for good 20Hz response
For 600 ohms (jurassic line input) 15H
For an output transformer, meant to be driven by an generic opamp (5532) you need 4H (5532 can drive as low as 150 ohms)
For a DOA (2520) you may go as low as 1.5H
The primary of a 70/100V line transformers has an inductance of about 200mH. It may seem very low, taken in isolation, but in the context of a power amp having an output impedance of <0.1 ohm, it's normal.
What is the impedance that drives the transformer when you do a sweep (quote: "yet when swept shows solid bottom response" )
The LC meter impedance, although not specified, is high.
A transformer's LF response depends not only on the absolute value of the inductance, it is the ratio of the primary impedance to the source impedance that matters.
For 200 ohm impedance (mic input) you need about 5H for good 20Hz response
For 600 ohms (jurassic line input) 15H
For an output transformer, meant to be driven by an generic opamp (5532) you need 4H (5532 can drive as low as 150 ohms)
For a DOA (2520) you may go as low as 1.5H
The primary of a 70/100V line transformers has an inductance of about 200mH. It may seem very low, taken in isolation, but in the context of a power amp having an output impedance of <0.1 ohm, it's normal.
what do you mean by response?
-3db @ 10hz?
distortion is the main problem at the low end due to increased excitation current ,
-3db would be saved for the high end unless you are drawing consumer graphs,
Gen Rad has an LCR meter with a jack for external osc so you can plot L vs F and L vs db,
power outputs can be tested by injecting a signal into the 6l6 inverter if you want high level testing,best way is to listen since the speaker and cabinet are usually non linear,
absolute value?
never seen negative henries unless the meter was weird,
L = V / (di/dt). The most probable thing that happened was that you (or your testing equipment) set something up backward involving a diode, or perhaps it needs to be re-calibrated.
Physically, V = μ * dФ/dt by Faraday's law, and Ф is proportional to i by Ampere's law, so anytime di/dt is negative, dФ/dt should also be negative and therefore V should be negative, so L should always be a positive number.
http://answers.yahoo.com/question/index?qid=20100519114124AA9Y9Qt
did you really think i was tht smart?
i talk so much shit i gots halitosis,
-3db @ 10hz?
distortion is the main problem at the low end due to increased excitation current ,
-3db would be saved for the high end unless you are drawing consumer graphs,
Gen Rad has an LCR meter with a jack for external osc so you can plot L vs F and L vs db,
power outputs can be tested by injecting a signal into the 6l6 inverter if you want high level testing,best way is to listen since the speaker and cabinet are usually non linear,
absolute value?
never seen negative henries unless the meter was weird,
L = V / (di/dt). The most probable thing that happened was that you (or your testing equipment) set something up backward involving a diode, or perhaps it needs to be re-calibrated.
Physically, V = μ * dФ/dt by Faraday's law, and Ф is proportional to i by Ampere's law, so anytime di/dt is negative, dФ/dt should also be negative and therefore V should be negative, so L should always be a positive number.
http://answers.yahoo.com/question/index?qid=20100519114124AA9Y9Qt
did you really think i was tht smart?
i talk so much shit i gots halitosis,
The sweep was actually done by Doug so I would assume source was set to rated pri Z = 600r.
My handheld gave reading of 4.5 and 5H @ 120Hz for each of a pair - this being on a "600r" pri.
The sweep Doug did looked as good or better on the bottom than, say, a UTC A-10 which will give a meter reading of 15-18H on the 500/600 tap.
use to get about 4 henries on each UTC A-10 or A-100X pri, combined for 16 henries minus any losses when linking the two coils,
largest pri inductance i ever saw on a 600 winding was 45 henries on the WE 111C, thus the killer bass, the core is crazy, must weigh about 2 lbs,
largest pri inductance i ever saw on a 600 winding was 45 henries on the WE 111C, thus the killer bass, the core is crazy, must weigh about 2 lbs,
Which would be correct for jurassic impedance matching but not for today's bridging. Today, the highest source impedance in professional environment is microphones and passive DI boxes; all other sources are typically 50-120 ohms.lassoharp said:
Is it possible that the sweep has been made at such a level that the UTC started to saturate? Has it been made as a direct compare, or is it possible something has changed in the test set-up?
some possible explanations for the low readings,
initial perm is lower than relative or max perm,
initial is close to the zero flux point on the BH curve,
relative perm is used in most calculations if at all,
it is the perm of the material related (thus the term relative) to air.
but relative perm can be different depending on who you ask,
this is because the slope of the B-H curve varies, and relative perm can be taken anywhere between u initial and u max.
so what is the relative perm of the UTC A-10 core material?
here is a spreadsheet on the UTC 31 UI lam, which was used in 90 percent of their transformers in various alloys
from the older Black A-10 to the newest blu-gray, UTC used between 436 and 449 turns of pri wire on each pri coil,
so if i inject 446 turns into the sheet, it tells me that according to the mag metals spec sheet, i have a perm of 30,000. but super 80 has a relative perm around 80,000,
so i am thinking the handheld meter did not apply a large enough voltage to really get the B-H curve going,
the upper formulas are for minimum turns required before saturation occurs for a given voltage on the core in question.
remember, when you series the two UTC primaries, henries should quadruple unless you made your measurements from 1 to 6 with 3 and 4 connected.
values on this sheet correspond to a 10 hz signal,
Turns vs Voltage @ Sat:
lets say you have a 1 volt signal coming into the transformer, E will equal 1, so for the 31DU core, you will need 1 times 496 = 496 turns or more to avoid saturation. (for 10 hz, superperm 80)
see that you need less turns to avoid sat as the alloy tends toward 4 percent silicon,
initial perm is lower than relative or max perm,
initial is close to the zero flux point on the BH curve,
relative perm is used in most calculations if at all,
it is the perm of the material related (thus the term relative) to air.
but relative perm can be different depending on who you ask,
this is because the slope of the B-H curve varies, and relative perm can be taken anywhere between u initial and u max.
so what is the relative perm of the UTC A-10 core material?
here is a spreadsheet on the UTC 31 UI lam, which was used in 90 percent of their transformers in various alloys
from the older Black A-10 to the newest blu-gray, UTC used between 436 and 449 turns of pri wire on each pri coil,
so if i inject 446 turns into the sheet, it tells me that according to the mag metals spec sheet, i have a perm of 30,000. but super 80 has a relative perm around 80,000,
so i am thinking the handheld meter did not apply a large enough voltage to really get the B-H curve going,
the upper formulas are for minimum turns required before saturation occurs for a given voltage on the core in question.
remember, when you series the two UTC primaries, henries should quadruple unless you made your measurements from 1 to 6 with 3 and 4 connected.
values on this sheet correspond to a 10 hz signal,
Turns vs Voltage @ Sat:
lets say you have a 1 volt signal coming into the transformer, E will equal 1, so for the 31DU core, you will need 1 times 496 = 496 turns or more to avoid saturation. (for 10 hz, superperm 80)
see that you need less turns to avoid sat as the alloy tends toward 4 percent silicon,
Attachments
I was only able to directly compare the A-10 and the other transmitter input transformer with the handheld meter, so, just the basic L readings at a single frequency. Doug may have some A-10 sweeps on file with rel test sig levels, source Z, etc. CJ have any for the A-10?
I will measure the outptut level of my meter across different loads just for reference. Gives about 1/2Vrms across DMM probes. I'm not sure if it uses a set internal Z to generate this across or not.
abbey road d enfer said:
Audio Precision set for 600 ohm source....
abbey road d enfer said:
....which is correct (JFCfGs) when testing jurassic era iron that assumed matching conditions.
Bridging is a backwards red herring in this line of questioning.
Anyway, the interpretation you have is backwards; what lassoharp said was the iron doesn't look so great with the handheld LCR meter tested open at 120 Hz, but looks fine on my AP with a 600 source into a 100K load.
The transformer in question is the input from an RCA AM broadcast transmitter; certainly a high level line to grid situation, if ever there was one. You would expect it to be driven directly from a BA-6 or other similar power amp grade limiter, and a handheld LCR meter or AP test set is probably not capable of blowing it up.
To editorialize, that's the grand improvement with using jurassic era iron when building jurassic era circuits; iron that actually performs properly under matching conditions. Modern iron is generally built with the assumption of bridging conditions, and you can't assume it can be stuck in a 1940's style zero feedback circuit and give acceptable response. I've seen 1970's broadcast/recording grade 10K:10K inputs that didn't give acceptable bottom end until driven with a 100 ohm source and a fixed 10K resistor on the secondary, and that's not in the available specifications. I've seen others that, on the surface, appear to be interchangeable with the one just mentioned, that work just fine for matching conditions, such that one might use them as interstage transformers in a tube circuit.
I don't think I've got it backwards. Lassoharp wrote: "high level input transformer from radio transmitter that gives a relatively paltry L reading using a handheld meter, yet when swept shows solid bottom response" then compared it to "UTC A-10 which will give a meter reading of 15-18H on the 500/600 tap" but doesn't fare as well on the sweep.emrr said:
That's in apparent contradiction, but since you mention the origin of this transformer, it is clear that it is intended for line level operation, so my analysis is that its relatively low inductance doesn't change as much with level as for the microphone input transformer, resulting in better LF performance at the considered operating level.
That's why I asked the measurements conditions, impedance and level.
Iron (including mu-metal bla bla) transformers without an air gap have an inductance that varies with level. I post examples earlier on this thread.
Williamson pontificates on this in his famous PA articles but it holds with small signal stuff too.
Williamson pontificates on this in his famous PA articles but it holds with small signal stuff too.
Permeability is denoted by small u.
notice that u=B/H (perm formula) is also the rise vs run of the BH curve,
so Permeability is the slope of the BH curve at whatever point you take it.
if you find the slope at the steepest part of the curve, this will be u max.
if you find the slope near the zero crossing of the graph, that will be u-i, initial
now if the LCR meter only has enuff juice to barely excite the coil, you will have a BH curve generated that only represents u-i.
since part of the current used to create H will be spent on exciting the core past it's losses, B/H will be flatter which means less slope=less perm=less inductance.
if the LCR meter has enuff juice to provide the excitation current and then some, then you will have a steeper BH, a steeper slope=more perm= more inductance.
so if you barely excite the xfmr, you will be getting a lower inductance reading due to the less steep BH curve.
nickel cores are easier to excite than silicon steel, so you will have more variation with level with silicon
Silicon Steels have a wide range of permeabilities depending on how hard they are driven (also known as amplitude permeability
this is because u-i for silicon is about 350, but u max is 10,000
50/50 alloy has u-i @ 3,000, and u-max @ 20,000, less delta between the 2,
80 ni has u-i @ 20,000, and u-max @ 40,000,
the less distance between u-i and u-max, the less error from different levels used to check L.
so you need a meter matched to the level of the xfmr you are checking to get a good reading for L.
now a gapped transformer or choke will have a constant perm/inductance over a wide range of levels but will take more excitation current from the LCR meter.
now if you read all that, you deserve a break today, so who shot JR?
here is a highlights clip, count the times he says drunk, when hagman used to drink 4 bottles of champagne a day, wtf, over?
http://www.youtube.com/watch?v=GZV3365a7Ew
notice that u=B/H (perm formula) is also the rise vs run of the BH curve,
so Permeability is the slope of the BH curve at whatever point you take it.
if you find the slope at the steepest part of the curve, this will be u max.
if you find the slope near the zero crossing of the graph, that will be u-i, initial
now if the LCR meter only has enuff juice to barely excite the coil, you will have a BH curve generated that only represents u-i.
since part of the current used to create H will be spent on exciting the core past it's losses, B/H will be flatter which means less slope=less perm=less inductance.
if the LCR meter has enuff juice to provide the excitation current and then some, then you will have a steeper BH, a steeper slope=more perm= more inductance.
so if you barely excite the xfmr, you will be getting a lower inductance reading due to the less steep BH curve.
nickel cores are easier to excite than silicon steel, so you will have more variation with level with silicon
Silicon Steels have a wide range of permeabilities depending on how hard they are driven (also known as amplitude permeability
this is because u-i for silicon is about 350, but u max is 10,000
50/50 alloy has u-i @ 3,000, and u-max @ 20,000, less delta between the 2,
80 ni has u-i @ 20,000, and u-max @ 40,000,
the less distance between u-i and u-max, the less error from different levels used to check L.
so you need a meter matched to the level of the xfmr you are checking to get a good reading for L.
now a gapped transformer or choke will have a constant perm/inductance over a wide range of levels but will take more excitation current from the LCR meter.
now if you read all that, you deserve a break today, so who shot JR?
here is a highlights clip, count the times he says drunk, when hagman used to drink 4 bottles of champagne a day, wtf, over?
http://www.youtube.com/watch?v=GZV3365a7Ew
