earthsled
Well-known member
I assume this means the the A15HP was intended for a system with matching impedance (150R mic to a 150R input). What could be done to modify the circuit for bridging impedance (150R mic to a 1K5 input)?
Thanks!
Balanced High Pass Filter
- Thread starter earthsled
- Start date
Help Support GroupDIY Audio Forum:
I assume this means the the A15HP was intended for a system with matching impedance (150R mic to a 150R input). What could be done to modify the circuit for bridging impedance (150R mic to a 1K5 input)?
Thanks!
Yes, loaded with 150R, the -3dB point is at 100Hz. The slope is about 6dB/oct and the in-band signal is attenuated by 6dB though.
Not a very good application, particularly considering this filter will be used very often on dynamics, who don't like to be so heavily loaded.
Which shows that, although the brochure is technically correct, it doesn't represent the product well.
Not a very good application, particularly considering this filter will be used very often on dynamics, who don't like to be so heavily loaded.
Which shows that, although the brochure is technically correct, it doesn't represent the product well.
earthsled
Well-known member
Thanks Abbey! The green graph looks like a good/practical slope for a high-pass. I suspect the mic pre's input transformer would create a parallel inductance to consider (reducing the overall inductance of the filter) -- or did you account for this in your model?
No, I haven't. You're right, having a transformer-based input changes the behaviour, generally increasing the -3dB frequency somewhat. That's the reason you can't design such a filter with the guarantee that it will provide the same performance with all sorts of mics and pres. You may find combinations that would would noticeably better than others.earthsled said:
earthsled
Well-known member
Indeed it's true. I just bought two of these for testing (catalog# 273-1380). One measures 0.688H and the other is 0.660H -- and a bargain at $2.69 each!
Wondering about two things...
The 2K7 resistors shown in parallel with the caps are to allow phantom power to pass, so for some they may not be needed. How do these resistors effect the impedance / frequency response of the filter, or do they?
The schematic shows polarized caps being used. I would guess tantalum would be a good choice given the small form factor. Is there any disadvantage to using polarized caps vs. non-polar or bi-polar types?
Thanks!
Without the resistors, there is a slight hump in the frequency response at ca. 150Hzearthsled said:
Never use tant caps in a position where ther's a risk of reverse voltage; that is very much the case here.
Yes, definitely. The distortion phenomenon in electrolytic caps depends directly on the voltage that is applied to it, that's why they can be used as coupling caps (DC blocking) but on the condition that they are calculated for a much lower -3dB frequency than needed. But here, as they are part of the filter, the caps are submitted to a significant portion of the input voltage. If it is typical mic voltage (10-100mV) it is acceptable. Ideally, you should use film capacitors here.
earthsled
Well-known member
Interesting stuff! 
Assuming film caps are too large, is there any advantage to using non-polar electrolytics, or should I expect the same distortion?
Attached is a test I made using a Lundahl LL1576 input transformer strapped for 1:7 operation. I connected a 100K resistor across the secondary so that the reflected input impedance should be about 2K. The output impedance of my signal generator is set at 200R to simulate a mic's output.
For these graphs, I omitted the 2K7 resistors. As mentioned above, I'm using the primary of the Radio Shack transformer for a little less than 0.7H inductance. The blue graph is using 2uF caps, while the green is using 10uF caps.
It appears 10uF caps produce a -3dB cutoff around 80Hz, while 2uF caps produce closer to 350Hz and both slopes are near -6dB per octave. So, I suppose 10uF will work best for me, but I was hoping for a steeper slope. Is this achievable by the use of different values, or would I need additional components?
Assuming film caps are too large, is there any advantage to using non-polar electrolytics, or should I expect the same distortion?
Attached is a test I made using a Lundahl LL1576 input transformer strapped for 1:7 operation. I connected a 100K resistor across the secondary so that the reflected input impedance should be about 2K. The output impedance of my signal generator is set at 200R to simulate a mic's output.
For these graphs, I omitted the 2K7 resistors. As mentioned above, I'm using the primary of the Radio Shack transformer for a little less than 0.7H inductance. The blue graph is using 2uF caps, while the green is using 10uF caps.
It appears 10uF caps produce a -3dB cutoff around 80Hz, while 2uF caps produce closer to 350Hz and both slopes are near -6dB per octave. So, I suppose 10uF will work best for me, but I was hoping for a steeper slope. Is this achievable by the use of different values, or would I need additional components?
Attachments
> I was hoping for a steeper slope.
Slope over the first 10dB is 10db/octave. This-many parts will ultimately approach 12db/oct, 2nd-order, actually never reaching 12db/oct.
Slope is as good as it gets.
Peaking at corner is -3db. This is Butterworth, an excellent filter when you have no spectific goal. You can also Bessel, which gives less phase-shift but a very broad-rounded knee, you can Tscheby which can give a bump and a short steep slope but wild phase and usually needs considerable latitude (large impedance difference or active gain).
I think it does not get better than this; or insignificantly better with extreme effort.
WHY are you high-passing at the mike input? The only reason I know is HUGE subsonics over weak speech/music. Booming trucks or blowers which OVERload the mike preamp before the desired signal comes up. Futher along the signal chain, filtering is easier. With the advent of PC sound edition, I never high-passed the recording, I fixed it in the PC (and I have run some STEEP digital hi-passes.... I hate subsonics).
Slope over the first 10dB is 10db/octave. This-many parts will ultimately approach 12db/oct, 2nd-order, actually never reaching 12db/oct.
Slope is as good as it gets.
Peaking at corner is -3db. This is Butterworth, an excellent filter when you have no spectific goal. You can also Bessel, which gives less phase-shift but a very broad-rounded knee, you can Tscheby which can give a bump and a short steep slope but wild phase and usually needs considerable latitude (large impedance difference or active gain).
I think it does not get better than this; or insignificantly better with extreme effort.
WHY are you high-passing at the mike input? The only reason I know is HUGE subsonics over weak speech/music. Booming trucks or blowers which OVERload the mike preamp before the desired signal comes up. Futher along the signal chain, filtering is easier. With the advent of PC sound edition, I never high-passed the recording, I fixed it in the PC (and I have run some STEEP digital hi-passes.... I hate subsonics).
> 1272 clone, followed by an LA-2A clone, then into an A to D
Ah.
Move away from the bass and drums.
R-C hi-pass in the LA2 input network, between input transformer and 39K(?) resistor. Unless you work under a dance-studio, 6db/oct slope ought to reduce duck-under good enough. 12db/oct would suggest a cathode-follower and Sallen Key filter, the unity-gain variant. If your thunder is worse, since you are adding a tube (or opamp) and dual is as easy as single, 30db/oct is possible (though very fussy).
Ah.
Move away from the bass and drums.
R-C hi-pass in the LA2 input network, between input transformer and 39K(?) resistor. Unless you work under a dance-studio, 6db/oct slope ought to reduce duck-under good enough. 12db/oct would suggest a cathode-follower and Sallen Key filter, the unity-gain variant. If your thunder is worse, since you are adding a tube (or opamp) and dual is as easy as single, 30db/oct is possible (though very fussy).
earthsled
Well-known member
I've found that Mouser has some transformers that are very close to the Radio Shack 273-1380.
42TL013-RC (1K:8, 75mW)
42TM013-RC (1K:8, 200mW)
42TU013-RC (1K:8, 400mW)
The 200mW version has almost identical dimensions to the Radio Shack model, except that it is PC-mount.
Here's my question: Is there formula that can be used to estimate the primary inductance of an off-the-shelf transformer?
Thanks!
42TL013-RC (1K:8, 75mW)
42TM013-RC (1K:8, 200mW)
42TU013-RC (1K:8, 400mW)
The 200mW version has almost identical dimensions to the Radio Shack model, except that it is PC-mount.
Here's my question: Is there formula that can be used to estimate the primary inductance of an off-the-shelf transformer?
Thanks!
Considering one the most important data of any wound component is its windings inductance, isn't it somewhat curious no xfmr manufacturer publishes this information?
> no xfmr manufacturer publishes this information?
Hammond 125ESE says 9.58 H (and 90.6 Ohms).
http://hammondmfg.com/125SE.htm
DC current not stated (and it matters).
Also interesting that the series claims +/-1db down to 100Hz, and a 10K impedance choice, but 9.58H at 100Hz is only 6K impedance.
My own tests say the claims are fairly close: working near 10K load I had few-db down at 150Hz with 50mA flowing.
> no xfmr manufacturer publishes this information?
It is in general not easy to spec or confirm such numbers, especially on non-gapped cores. The very low flux inductance will be one thing, medium-flux inductance will be higher, and then it will fall at high-flux. It will also be higher below 200Hz and lower above 400Hz due to eddy-currents throwing the flux out to the skin of the iron.
The japan-radio transformers are so loosely assembled that they may act semi-gapped, without a large rise from low to medium flux; they sure do drop if you approach the maximum V/F limit.
Hammond 125ESE says 9.58 H (and 90.6 Ohms).
http://hammondmfg.com/125SE.htm
DC current not stated (and it matters).
Also interesting that the series claims +/-1db down to 100Hz, and a 10K impedance choice, but 9.58H at 100Hz is only 6K impedance.
My own tests say the claims are fairly close: working near 10K load I had few-db down at 150Hz with 50mA flowing.
> no xfmr manufacturer publishes this information?
It is in general not easy to spec or confirm such numbers, especially on non-gapped cores. The very low flux inductance will be one thing, medium-flux inductance will be higher, and then it will fall at high-flux. It will also be higher below 200Hz and lower above 400Hz due to eddy-currents throwing the flux out to the skin of the iron.
The japan-radio transformers are so loosely assembled that they may act semi-gapped, without a large rise from low to medium flux; they sure do drop if you approach the maximum V/F limit.
OK, I hadn't seen that. They publish just this one figure when there are 7 different models of transformers. I believe it just shows people do not care about this parameter, they just want to be told it's a 600 ohms, 10k or whatever primary.PRR said:
I suspect this Hammond xfmr has the same problem, and yet, someone had a revelation and thought that it may be useful information, even if the torerance is -50/+100%. Most inductors have the same problem, but manufacturers sure do indicate some inductance value.
I understand that most buyers are satisfied with "Primary impedance range from 2,500 to 10,000 Ohms. Secondary impedance range from 4 to 32 Ohms." but when I shop for a transformer, I like to know more. I think it shouldn't be too difficult to have the inductance listed somewhere, for those who are just a little more curious.
CJ
Well-known member
inductance is a by-product of the design process,
for pwr transformers, you pick a core weight that corresponds to the power you are handling, then you find a wire that will handle the current, then you calculate how many turns you can squeeze into the window and you are done.
QC will check turns ratio, but has no use for inductance ratings.
what do you do when designing an audio transformer?
you find your voltage level, determine what size core you need to keep the flux down low enough to meet your distortion goal, then you wind it for whatever impedance you want to bridge,
so inductance is actually built into the frequency response graph,
some companies have henries written into the catalog, but most do not.
you want a transformer that the 1 db down from 10 to 20 k hertz, not a transformer with 10 henries,
for pwr transformers, you pick a core weight that corresponds to the power you are handling, then you find a wire that will handle the current, then you calculate how many turns you can squeeze into the window and you are done.
QC will check turns ratio, but has no use for inductance ratings.
what do you do when designing an audio transformer?
you find your voltage level, determine what size core you need to keep the flux down low enough to meet your distortion goal, then you wind it for whatever impedance you want to bridge,
so inductance is actually built into the frequency response graph,
some companies have henries written into the catalog, but most do not.
you want a transformer that the 1 db down from 10 to 20 k hertz, not a transformer with 10 henries,
Please allow me to disagree, especially when designing inductors.CJ said:
PT industry is not concerned by frequency response and source impedance.
That's what one does when designing transformers (or inductances) with a coobook. Designing for power handling but neglecting the frequency response may lead to a transformer capable of the requested power but limited to 60 Hz LF response.
So, has someone actually cared about frequency response or not?
Maybe you, not me, I want as much info as I need. Knowing the inductance is essential in optimizing the circuitry around it.
CJ
Well-known member
another reason is that it is easy to fudge an inductance spec,
if company A says that there pri inductance is 10 henries,
then company B can say 20 henries because they measured at a distorted flux level or at a different frequency.
the spec is a very loose way of looking at transformers and inductors, the transformer geeks all recognize this, so they choose to avoid talking about an very inexact spec.
hope all our members are doing ok,
were are all ok, right?
if company A says that there pri inductance is 10 henries,
then company B can say 20 henries because they measured at a distorted flux level or at a different frequency.
the spec is a very loose way of looking at transformers and inductors, the transformer geeks all recognize this, so they choose to avoid talking about an very inexact spec.
hope all our members are doing ok,
were are all ok, right?
earthsled
Well-known member
It looks like the short answer is: the inductance of a transformer can not be calculated based on other known factors.
This leads me to presume PRR must have measured the Radio Shack xfrmr in the past.
I've learned that inductors can be calculated based on core dimensions, wire dimensions, number of turns, core materials, etc. So, I would suppose two transformers sharing the same ratio and form-factor may have similar inductances. Still, the only way to be certain would be to measure them.
This leads me to presume PRR must have measured the Radio Shack xfrmr in the past.
I've learned that inductors can be calculated based on core dimensions, wire dimensions, number of turns, core materials, etc. So, I would suppose two transformers sharing the same ratio and form-factor may have similar inductances. Still, the only way to be certain would be to measure them.
> PT industry is not concerned by frequency response and source impedance.
They are, but in different ways.
Since primary inductance is never infinite, there is alwasy some reactive current. The current is "imaginary", and can be tuned-out with capacitance, but the I2R losses in the coil and line are very real.
In effect, they want "bass response better than 0.5db at 60Hz". Instead they call it magnetizing loss.
It seems to work-out (perhaps by intent) that 60Hz above 1KVA on anything better than stove-iron, the reactive losses are "small" relative to total power or to full-load losses. Good enough for commercial loads.
OTOH, run the mains up Wild Goose Road, houses 500 feet apart, some of them summer-only: you have a large number of small 10KVA transformers sucking energy 24/7 and relatively small billable load. Poor profit.
This is an "industry" problem. The lineman rarely computes the losses, but the utility industry and the iron-mongers have gradually improved irons. Standby losses are a key specification for some utility designers.
That's big iron. Little 10VA wall-warts sometimes have magnetizing current greater than load current. The first-cost is cheaper that way. The running cost may exceed the first-cost, but are still "small", and small-stuff buyers don't look at running-cost.
> presume PRR must have measured the Radio Shack xfrmr
Nope. I know it is good, as "1K", down to 300Hz-150Hz. 1H at 160Hz is 1K. Desired value in the Shure filter is 0.7H. It's ballpark, VERY cheap/available, and any "error" can be skewed by capacitor changes (much easier than finding other inductor). I also know it starts to grossly distort in a 9V pocket radio, over 10V across the primary, so it "may" be low-THD at mike-level. It's not shielded, but you will learn not to lay it on power transformers or other hotspots.
They are, but in different ways.
Since primary inductance is never infinite, there is alwasy some reactive current. The current is "imaginary", and can be tuned-out with capacitance, but the I2R losses in the coil and line are very real.
In effect, they want "bass response better than 0.5db at 60Hz". Instead they call it magnetizing loss.
It seems to work-out (perhaps by intent) that 60Hz above 1KVA on anything better than stove-iron, the reactive losses are "small" relative to total power or to full-load losses. Good enough for commercial loads.
OTOH, run the mains up Wild Goose Road, houses 500 feet apart, some of them summer-only: you have a large number of small 10KVA transformers sucking energy 24/7 and relatively small billable load. Poor profit.
This is an "industry" problem. The lineman rarely computes the losses, but the utility industry and the iron-mongers have gradually improved irons. Standby losses are a key specification for some utility designers.
That's big iron. Little 10VA wall-warts sometimes have magnetizing current greater than load current. The first-cost is cheaper that way. The running cost may exceed the first-cost, but are still "small", and small-stuff buyers don't look at running-cost.
> presume PRR must have measured the Radio Shack xfrmr
Nope. I know it is good, as "1K", down to 300Hz-150Hz. 1H at 160Hz is 1K. Desired value in the Shure filter is 0.7H. It's ballpark, VERY cheap/available, and any "error" can be skewed by capacitor changes (much easier than finding other inductor). I also know it starts to grossly distort in a 9V pocket radio, over 10V across the primary, so it "may" be low-THD at mike-level. It's not shielded, but you will learn not to lay it on power transformers or other hotspots.
earthsled
Well-known member
I see. Your estimate was based on the frequency response of the transformer. Perhaps a transformer with the same specs for frequency response would have similar inductance?
