That is exactly the explanation I reckon for lassoharp's apparent contradiction. The "visible" effects of core saturation vary very much with the primary damping (DCr + source Z).ricardo said:
Audio Transformer Inductance
- Thread starter ruffrecords
- Start date
Help Support GroupDIY Audio Forum:
CJ
Well-known member
here is a graph showing how permeability varies with flux level.
this is for a non gapped grain oriented steel core,
the top graph is with no DC on the core, the family of curves are for various amounts of dc thru the coil.
so you can see how inductance changes with level because perm changes with level.
this graph represents the first derivative of the BH curve,
this is for a non gapped grain oriented steel core,
the top graph is with no DC on the core, the family of curves are for various amounts of dc thru the coil.
so you can see how inductance changes with level because perm changes with level.
this graph represents the first derivative of the BH curve,
Attachments
UTC LS-51 (mid period grey painted badged variant) on an inductance meter with a 120 Hz test, no load, and saw:
2487 Henry on the 30K winding, DCR is 2015.
34.57 Henry on the 500 winding, DCR is 46.
in comparison, jackies tested an A-26 at 100 Hz and saw:
1276 Henry on the 30K winding
21 Henry on the 500 winding
Interesting point, the fudge factor between the lines on paper. I will argue you can't buy transformers like the LS-51 from anyone these days.
The Sowter 8344E used for Fairchild 660/670 clones is meant for PP 60K:600. I measured two:
24H, 492 DCR : 1H, 5R5 DCR
38H, 524 DCR : 1H, 5R5 DCR
As PRR and CJ have both said before, there are positive aspects to higher DCR in windings intended for tube use, particularly with limiters.
The Sowter is steel core, the UTC has high nickel content. Totally apples and oranges, in every regard.
10 years ago, when the LS-51 could still be ordered on a custom basis, it was nearly a $400 transformer.
2487 Henry on the 30K winding, DCR is 2015.
34.57 Henry on the 500 winding, DCR is 46.
in comparison, jackies tested an A-26 at 100 Hz and saw:
1276 Henry on the 30K winding
21 Henry on the 500 winding
Interesting point, the fudge factor between the lines on paper. I will argue you can't buy transformers like the LS-51 from anyone these days.
The Sowter 8344E used for Fairchild 660/670 clones is meant for PP 60K:600. I measured two:
24H, 492 DCR : 1H, 5R5 DCR
38H, 524 DCR : 1H, 5R5 DCR
As PRR and CJ have both said before, there are positive aspects to higher DCR in windings intended for tube use, particularly with limiters.
The Sowter is steel core, the UTC has high nickel content. Totally apples and oranges, in every regard.
10 years ago, when the LS-51 could still be ordered on a custom basis, it was nearly a $400 transformer.
This is very strange; that suggests the -3dB LF corner at about 200Hz.emrr said:
Have you actually the possibility to check them in circuit, or measure the inductance with a more traditional set-up (bridge, resonant circuit or attenuator)? Both the primary and secondary values look out by a factor of 5-10. It may be interesting to have Brian Sowter's opinion. He can be contacted at: [email protected]
Not enough turns???!... 
That's how I feel. Yet the DCR is right in line with the spec Brian passed on via email when I purchased. It implies much larger wire at very least. Brian quoted 62H, and I won't pretend my equipment is able to test as well as his.
A vintage Ferranti 50K:500 PP output transformer used in similar fashion has DCR of 7K:36; no inductance measurement on that one yet.
A vintage Ferranti 50K:500 PP output transformer used in similar fashion has DCR of 7K:36; no inductance measurement on that one yet.
ruffrecords said:
Ian:
Attached is a spreadsheet containing the jackies measurements. I suggest you track the person making the measurement and the equipment, in order to make the data comparable from a procedure/equipment point of view.
The spreadsheet is in tab delimited format which should import nicely into a spreadsheet (site does not allow the xls formats).
Bruce
Very interesting indeed!
There are many inconsistencies. One factor that can't be questioned in the mfgr's specs is the ratio.
In the case of e.g. LL1636, the ratio is 1:10, so the inductance ratio should be 1:100; the measurements show a ratio of 1:52.
The JLM1:4 shows the same discrepancy, with an impedance ratio of 1:27, much more than the expected 1:16.
The Sowter 3195 however seems to measure as predicted, with a measured ratio of 1:56, close enough to the theoretical 1:49.
The extremely high values for some of the UTC's, seem to indicate that the LCR meter has a large positive error when presented with high value inductors.
I've checked the documentation for the MS5308. It seems that everything is more or less autoranging, meaning that test conditions vary from one test to the other.
Having owned a quite sophisticated LCR meter (Racal-Dana 9343), I quickly found out that it could not be expected to read reliable numeric values of iron or ferrite cored inductors. So i used a different test set (frequency response of an RLC circuit) for assessing the real value and used the Racal Dana meter for production, where the actual inductors were compared to a golden sample.
There are many inconsistencies. One factor that can't be questioned in the mfgr's specs is the ratio.
In the case of e.g. LL1636, the ratio is 1:10, so the inductance ratio should be 1:100; the measurements show a ratio of 1:52.
The JLM1:4 shows the same discrepancy, with an impedance ratio of 1:27, much more than the expected 1:16.
The Sowter 3195 however seems to measure as predicted, with a measured ratio of 1:56, close enough to the theoretical 1:49.
The extremely high values for some of the UTC's, seem to indicate that the LCR meter has a large positive error when presented with high value inductors.
I've checked the documentation for the MS5308. It seems that everything is more or less autoranging, meaning that test conditions vary from one test to the other.
Having owned a quite sophisticated LCR meter (Racal-Dana 9343), I quickly found out that it could not be expected to read reliable numeric values of iron or ferrite cored inductors. So i used a different test set (frequency response of an RLC circuit) for assessing the real value and used the Racal Dana meter for production, where the actual inductors were compared to a golden sample.
abbey road d enfer said:
How do you mean? The frequency is selected, and I measured the voltage applied (can't recall how much exactly) and it stays the same. What other test conditions are we talking about besides the frequency and signal level?
What voltage did you measure? The voltage under load cannot be constant. Most LCR meters operate under more or less constant current, making the measurements of very low or very high values prone to error.jackies said:
abbey road d enfer said:
I did a little data analysis too, and I did not see the result that way. You will notice that on those measurements, both the primary and the secondary measure very high. In fact if you plot H/DCR on both primary and secondary and scattter plot them you get a pretty much straight line.
I was assuming those transformers differ due to core material (which can be dramatic) and wire size.
What I saw in this is that most audio transformers are about the same (small core, small wire, lots of turns, layered/shielded or rmultifilar), and that means that a reasonable guess at primary inductance can be made knowing only the DCR's. With the problem of core material lurking there.
I didn't add all the original measurements scattered through the thread, I think Ian has those. I would be good to see your calculations. Any chance we can get Ethan to allow XLS and XLSX files?
Still I wonder why the inductance figures seem to indicate these xfmrs are designed for 5Hz LF corner frequency.bruce0 said:
For example, the HA101 secondary 3575H computes at 450 kohms at 20Hz but it is rated at 120 kohms. It may be because the saturation magnetic force H is much lower in Hipermalloy, so they're taking a safety coefficient? They specify +16dBu on the primary. the inductance measurement should probably be done in the same conditions...
Most xfmr designers seem to make DCR about 1/10th of the rated impedance, which makes sense. But I wouldn't count on it to evaluate the inductance because the difference in LF corner frequency introduces a large variation. I've looked at the UTC specs; some xfmr's are designed for 20Hz, some for 30, some for 50 and some for 10Hz.
CJ
Well-known member
those dual coil UTC transformers always have mega turns,
8,000 on the pri of the A-26 = about 1500 Henries,
A-10, same thing,
LS-51 looks interesting,
all the UTC stuff uses the same lam,
so the LS series has the same lam as the A-series,
why so big? huge forged outer case and three more inside of that,
they just increase the lam stack and drop the turns for lower dcr,
do not take the specs you read on a xfmr too seriously as they are very loose,
at least on the stuff i have tested,
notice nobody goes down to 1 hz? always 5,
HP 200 has 5 hz for it's lowest frequency,
8,000 on the pri of the A-26 = about 1500 Henries,
A-10, same thing,
LS-51 looks interesting,
all the UTC stuff uses the same lam,
so the LS series has the same lam as the A-series,
why so big? huge forged outer case and three more inside of that,
they just increase the lam stack and drop the turns for lower dcr,
do not take the specs you read on a xfmr too seriously as they are very loose,
at least on the stuff i have tested,
notice nobody goes down to 1 hz? always 5,
HP 200 has 5 hz for it's lowest frequency,
Still haven't been able to test the Sowters, but here's another for the logs.
RCA 86 limiter, audio amp, interstage transformer. Note there is also a side chain interstage transformer which is different.
Inductance and AC ohm tests with hand-held Extech 380193 LCR meter
120 hz tests, series mode, unloaded
Primary
2 windings, each 470H, 116K AC ohm, 2660 DC resistance
in series as in circuit 1861H, 435k AC ohm, 5320 DC resistance
Secondary
932H, 224K AC ohm, 3790 DC resistance
RCA 86 limiter, audio amp, interstage transformer. Note there is also a side chain interstage transformer which is different.
Inductance and AC ohm tests with hand-held Extech 380193 LCR meter
120 hz tests, series mode, unloaded
Primary
2 windings, each 470H, 116K AC ohm, 2660 DC resistance
in series as in circuit 1861H, 435k AC ohm, 5320 DC resistance
Secondary
932H, 224K AC ohm, 3790 DC resistance
Here's some Altec 458A data.
Extech 380193 LCR meter, 120hz testing, series or parallel meter setting noted.
I may be trying to read individual winding C incorrectly, along with leakage L.
Input:
Pri series reading, 2 windings separate
6.75H, 2392 AC Z, 238 nfd, 46 dcr
6.82H, 2500 AC Z, 238 nfd, 44 dcr
Sec parallel reading
5185H, 707k AC Z, 0.323 nfd, 7k25 dcr
169 load on 150 pri 189k68 AC Z, 5490H, off scale cap
Pri shorted, 29k AC Z, 1935H, off scale cap (leakage L)
Output:
Pri parallel reading
307H, 52k7 AC Z, 5.49 nfd, 1020 dcr
620 load on sec 25k AC Z, 291H, 503 nfd
Sec shorted, 2k7 AC Z, 123H, cap off scale (leakage L)
Sec series reading, 2 windings separate
2.73H, 1010 AC Z, 780 nfd, 22 dcr
2.73H, 1033 AC Z, 764 nfd, 22 dcr
Full amp response plots for various loaded/unloaded conditions. I didn't look at tube current for best lows, just known good tubes. Unloaded input is MOTU converters as source, which is roughly 150 ohms and never happy with driving inductive sources at high frequencies. Loaded input is the same converters driving a typical 20dB U pad. Loaded output is 620R, unloaded is actually 10K converter input. Relative gain left intact in plots. In/out loaded looks broadest in response. Lack of resonant bumps is impressive.
Extech 380193 LCR meter, 120hz testing, series or parallel meter setting noted.
I may be trying to read individual winding C incorrectly, along with leakage L.
Input:
Pri series reading, 2 windings separate
6.75H, 2392 AC Z, 238 nfd, 46 dcr
6.82H, 2500 AC Z, 238 nfd, 44 dcr
Sec parallel reading
5185H, 707k AC Z, 0.323 nfd, 7k25 dcr
169 load on 150 pri 189k68 AC Z, 5490H, off scale cap
Pri shorted, 29k AC Z, 1935H, off scale cap (leakage L)
Output:
Pri parallel reading
307H, 52k7 AC Z, 5.49 nfd, 1020 dcr
620 load on sec 25k AC Z, 291H, 503 nfd
Sec shorted, 2k7 AC Z, 123H, cap off scale (leakage L)
Sec series reading, 2 windings separate
2.73H, 1010 AC Z, 780 nfd, 22 dcr
2.73H, 1033 AC Z, 764 nfd, 22 dcr
Full amp response plots for various loaded/unloaded conditions. I didn't look at tube current for best lows, just known good tubes. Unloaded input is MOTU converters as source, which is roughly 150 ohms and never happy with driving inductive sources at high frequencies. Loaded input is the same converters driving a typical 20dB U pad. Loaded output is 620R, unloaded is actually 10K converter input. Relative gain left intact in plots. In/out loaded looks broadest in response. Lack of resonant bumps is impressive.
