Hello. I would like to build a reamp for a friend.
I would like to find a transformer that works fine in the full 20Hz-20kHz bandwidth.

I prefer a cheaper transformer. But I want to be sure that I can find something that doesn't cut frequency.
I knew that the 42TM018 has a bandwidth of only 300Hz to 3.4kHz.
The Tamura MET-31 seems to have a bandwidth that start from 300Hz.

I watched this: https://www.youtube.com/watch?v=aC35Dic2cAc
In the description the transformer is this: VIGORTRONIX VTX-101-1604
https://it.farnell.com/vigortronix/vtx-101-1604/trasformatore-audio-1-1ct-600/dp/2723450
Datasheet: https://www.farnell.com/datasheets/4000531.pdf
The guy talks about 20Hz-20kHz and show some tests. While the specific of this transformer in the shop page, and in the datasheet, says it has a 60Hz like minimum frequency. So I not completely sure about this transformer.

There's this Edcor PC10K-10K: https://www.don-audio.com/Edcor-PC10K-10K
Datasheet: https://www.don-audio.com/mediafiles//datasheets/PC10K-10K.pdf
It seems a good one, a bit hard for me fit it without a PCB, and a bit expensive.

Neutrik NTE10-3
http://www.banzaimusic.com/neutrik-nte10-3.html
Datasheet: https://asset.conrad.com/media10/add/160267/c1/-/en/000515964DS02/datasheet-515964-neutrik-nte103-impedance-50000-primary-voltage-12-v-content-1-pcs.pdf
Cheap enough, but this one is not 1:1, and the application should be more than else for mic preamp. I have some doubt about this.

And the last, TY-250P: https://www.musikding.de/TY-250P_1
https://www.musikding.de/TY-250P_1
Datasheet: http://catalog.triadmagnetics.com/asset/ty-250p.pdf
This is cheap, 20Hz-20kHz, 1:1... almost too much perfect to be good.

Thanks for the help.

Quote from: Elijah-Baley on October 09, 2023, 06:31:04 AMI knew that the 42TM018 has a bandwidth of only 300Hz to 3.4kHz.
The Tamura MET-31 seems to have a bandwidth that start from 300Hz.

The 300-3.4K transformers are specified for " voice communications" purposes and for modems. However, they are still transformers.
In fact, they commonly exceed 20kHz high end; the manufacturers will only >guarantee< that they get over 3.4kHz. I've measured many of them.
The low end spec of a transformer is very loading and drive impedance sensitive. If you drive the primary from a much lower impedance than it's specified at (like, say, the output of an opamp or driver chip) then the low end rolloff gets extended. I tinkered opamp drive to a 42TM018 and measured -3db at 60-ish Hz. That's anemic for bass (low E at 42Hz), but not terrible. Take a look at the transformer coupled splitters at geofex.com. Also see:
https://www.diystompboxes.com/smfforum/index.php?topic=64924.20
====

February 27, 2008, 02:41:27 PM
Last Edit: February 27, 2008, 03:24:17 PM by R.G.

    Quote from: alfafalfa on February 27, 2008, 11:57:22 AM

    Auke in the datasheet of the Mouser Xformer 42TM018  it says that they have a freqency range of 300 Hz tot  3.4 KHz within 3 db  only !

    In my opinion it narrows down the performance too much especially in the low end.

That's a common mistake among people who don't understand transformers and datasheets, alf.

First about datasheets. Datasheets present the numbers that the manufacturer is willing to stand behind, in the sense that if his parts don't do at least that well, he'll give your money back. Some datasheets give min and max values of certain specifications, but many only give either a minimum (or maximum) and a typical value. For "typical" you should understand that the maker is saying "we think most of them are close to this. We hope the ones you get are close too. But we won't give your money back if it's not.

The specifications on bandwidth of transformers in general, and these in particular is a minimum of 300Hz to 3kHz. Why that? That's voice bandwidth on telephones. The actual transformers you get are almost certain to be better than that minimum - but how much? I did the obvious thing and measured it. Right off the shelf, the Xicon transformers I measured were about 150Hz to 22kHz within +/-3db. The manufacturer is obviously not going to go do work to make his products enough worse to make the specs come true, right?

Still, 150Hz is not all that good for guitar. So I used a trick. Transformer bandwidth is tremendously impedance sensitive. To get the best bandwidth in a passive situation, you simply have to drive it from the expected input impedance and load it to the expected output impedance. But I wasn't building a passive box, I was building a box to make the best use of cheap transformers.

So I took note of the fact that the low end of a transformer is limited by the core inductance eating up the input signal before it could be transformed. If you supply a primary from a quite low impedance, it can supply the extra current that the transformer core is trying to eat up, and still supply the signal being transformed too. In effect, the extra power extends the low end response on the secondary side.

In the circuit I set up, I ... measured... 60Hz to 22kHz, which I thought was good enough for rock 'n' roll.

====

If there's a bass roll-off, you could compensate with a boost in the driver stage.

It's going to be tricky to get something that gives a completely flat response when combined with the transformer, but that's not really the aim anyway, it it? It's more about "character" and not losing too much low end (or high end). And that should be do-able, with a bit of tweaking in the driver, even if the tranny isn't that great.

YMMV, I guess. *I'd* be happy is I suppose what I'm saying!

The frequency response spec of a transformer is measured with a source impedance and load impedance which match the specified primary and secondary impedances.   The source impedance is chosen to keep the insertion loss low eg. 2dB, 1dB, or 0.5dB.

If you drive the transformer with a lower source impedance you will lower the bass cut-off frequency.   There is no rule that you have drive the transformer with a match impedance.  It's been discussed on the forum quite a few times.

The problem with small transformers is they have higher winding resistance than larger transformers.   For a given transformer size the more you try to lower the bass cut-off the more the higher the winding resistance and at some point the insertion loss goes outside of what is acceptable - in terms of reasonable insertion loss.

The small Triad TY-250P transformer you are looking at has fairly high resistance.   The XICON transformers of a similar size have lower resistance, and as a result has a poorer low frequency spec.   To some degree construction can be a factor as well for low frequencies.

Another limit is the maximum input voltage for a given frequency.   The lower the frequency the lower the input voltage.  This is limited by physics and materials.

Physically large transformers will generally have a better lower frequency response and a higher maximum input voltage.

In short if you want good low frequency response you generally need a larger transformer.  If you look at DI transformers they tend to be 25mm cores to 40mm cores whereas those smaller transformers are only 19mm.   Some 600 ohm transformers can be used for DI if you drive them with a low impedance.

There's some info here,
https://sound-au.com/articles/audio-xfmrs.htm

The transformer here is 35mm x 29mm.

Look up some old posts on the forum.

What they said.

Also: 20cps-20kcps is a made-up spec. You do NOT have any 20Hz speakers. If your ears have spent time on a stage you don't have much 20kHz left; anyway the few instruments which do are annoying. 50-15Khz is enough, if smooth. 50-15k WAS the magic spec until like 1953, when someone (Fisher?) needed a "New for 1954!" blurb and noted that NFB amplifiers often beat this. High Fidelity ad-writers have not been entirely good for Music.

Even 50-15K and "cheaper" conflict.

Non-transformer differential line drivers and receivers exist.

Ok, thanks. There's a lot of theory. But I think I got the important part.

I had the idea that the cut-off depends from the impedance setting, so the transformer's specific need to be seen in the context. This is why the transformers used in the video (Vigortronix) works fine even at lower frequency, though the datasheet limit at 60Hz.
About this we can discuss on a schematic later, I guess, when I find the right transformer for my purpose (my frends's purpose): a more flat possible reamp. He used to play with guitar lower tuning, so it's important to have the same feeling when he plays guitar directly in the amp.
Even if I understand the 20Hz-20kHz thing is kind of theory.

This is a nice project: https://diy-fever.com/effects/lehle-splitter-re-amp/
The Lehle transformers is very good. The 42TM018 is not so bad as the specific said, but I would like something better.

I prefer to avoid to include circuits, especially if I have to correct transformer's issue.
Rather, I'm considering a buffer. (Is it necessary in front the transformer? Maybe I seen some schematic with buffer after the transformer... I'm not sure).

About the model of the transformer, I think the Vigortronix is a good choice, and the one I trust more, until I can get more about the other two. indeed, it is, with the Neutrik NTE10 and the TY-250P, in "small transformer family".

What about the "maximum input voltage for a given frequency"? When that is a problem?

Quote from: Elijah-Baley on October 10, 2023, 04:27:25 AMOk, thanks. There's a lot of theory. But I think I got the important part.

I had the idea that the cut-off depends from the impedance setting, so the transformer's specific need to be seen in the context. This is why the transformers used in the video (Vigortronix) works fine even at lower frequency, though the datasheet limit at 60Hz.
About this we can discuss on a schematic later, I guess, when I find the right transformer for my purpose (my frends's purpose): a more flat possible reamp. He used to play with guitar lower tuning, so it's important to have the same feeling when he plays guitar directly in the amp.
Even if I understand the 20Hz-20kHz thing is kind of theory.

As far as the video goes.  The circuit used is driving the transformer input directly from a low impedance source.   That will give a low frequency cut-off somewhat lower than what a manufacturer would specify (which is usually a matched source and load).  So you can't really compare the video results with the Vigortronics spec.  The video gets a -3dB point of about 10Hz but the spec is 60Hz; and even then the spec could -3dB @ 60Hz or -1dB at 60Hz, it's not clear.

The main point is the 10Hz is only possible if you drive the transformer with a low impedance.  That means a buffer.  And if you drive some of the other transformers with low impedance then the low-frequency point would be better than the specs.

QuoteThis is a nice project: https://diy-fever.com/effects/lehle-splitter-re-amp/
The Lehle transformers is very good. The 42TM018 is not so bad as the specific said, but I would like something better.

Yes, probably best to aim at something a little better.

QuoteI prefer to avoid to include circuits, especially if I have to correct transformer's issue.
Rather, I'm considering a buffer. (Is it necessary in front the transformer? Maybe I seen some schematic with buffer after the transformer... I'm not sure).

Generally you will need a buffer unless you get a special transformer designed for guitar/bass DI connections.  These transformers have much higher nominal impedances than the smaller ones in this thread.  Also somewhat more expensive.  Full passive is a whole different angle.

The set-up in the Vigortronix video is not set-up to accept a direct guitar input.  Moreover the input impedance is going to be in the order of 10k which will definitely load down a guitar pickup.

QuoteAbout the model of the transformer, I think the Vigortronix is a good choice, and the one I trust more, until I can get more about the other two. indeed, it is, with the Neutrik NTE10 and the TY-250P, in "small transformer family".

The thing that stands out to me about the Vigortronix is the core is probably only 16mm (EI16 core and 18mmx18mm with the plastic cover) whereas the Triad TY-250P is 19mm (EI19 core).   The TY-250P is spec'd at 20Hz whereas the Vigortronix is spec'd at 60Hz.   The larger core of the Triad seems desirable.

QuoteWhat about the "maximum input voltage for a given frequency"? When that is a problem?

If you go to 7:40 to 8:00 in the video
https://www.youtube.com/watch?v=aC35Dic2cAc

You can see that,
20Hz   3Vp-p   (1.1V rms) bad distortion
but at 30Hz the transformer is out of saturation
At 1kHz 13.6Vp-p (4.8V rms) clean

The voltage limit depends on frequency, in fact the voltage limit is proportional to frequency: the voltage limit will be near 6Vp-p s at 40Hz and 12Vp-p at 80Hz.   Each transformer will have a different limit at the same frequency.   It's not a simple matter to workout the voltage limit from the manufacturers specs.   The Triad transformer specifies a limit of 4.2V rms but no frequency is given (I could guess 20Hz but I have doubts) and we don't know how distorted it is compared to the Vigortronix.   Another spec the Triad gives is 20mW output power, which gives numbers in the order of the 4.2V rms.

The Mouser/Xicon transformers are quoted at 200mW, that's quite a high voltage, but the frequency where saturation/distortion occurs is in the 200Hz to 300Hz region.  So at 20Hz the voltage will be 10 times lower than the what is implied by that spec!

It's very easy to buy a transformer and test it.   However you can get caught out by not comparing apples to apples, or not understanding the true loading on the input signal - like the Vigortronix video.  It's fairy difficult to take manufacturers specs and make use of them.  Without measurements I could spend a few hours effectively designing a transformer which attempts to match-up the manufacturers specs in order to workout what the transformer will do.  Even then the estimates can have a large error.

Unfortunately the fine details of this stuff is overwhelming.

I have a few TY-250s lying around. Is there something I can breadboard and measure that would be helpful?

They sounded decent to my ear but I didn't push it hard or run any actual tests. The signal was also just single coil guitar so not a lot of lows to compare and a high preexisting noise floor.

Thank you, kenjib. If you can do it for me I'll appreciate very much.

I don't know your gear and tools, but I don't have anything but amps and guitars.
Actually, I might build it, but just my friend can test the reamp, and only with his ears.

Something I forgot to mention before is if you only want to use this to tap-off an amp output then it is possible to have a passive set-up.   

You need to place divider before the transformer.   The divider impedance needs to be low enough to ensure the source impedance driving the transformer is low and that ensures good bass response.

Next the amount of signal drop on the voltage divider is chosen so the transformer won't saturate when the amp is cranked.

---

I found this post,
https://www.talkbass.com/threads/an-inexpensive-audio-transformer-with-diy-projects.1113802/page-2
Which states,
"The TY250 has a measured primary inductance of 3.9H (we measured this for another project that we were working on)."

From that it gives some confidence that the 20Hz spec in the Triad datasheet is believable.   The test set-up would probably be 1k input source impedance and 1k load and 3.9H predicts a -3dB cut-off of 24Hz.

As mentioned earlier in the thread, the lower -3dB cut-off is lowered further when driven by a low source impedance.

The inductance of small transformers can vary with drive level, it can vary quite a bit.  Also the measured inductance can depend on the drive level of the LCR meter.   The varying inductance with level makes it difficult to compensate the bass drop of a transformer with an active bass-boost circuit.  I wouldn't be surprised if this why Rod Elliot get different inductance measurements on different LCR ranges.  He says the measurement is wrong but it could actually be a real effect,
https://sound-au.com/articles/audio-xfmrs.htm

---

Test results, probably with low impedance source and high impedance load.
https://www.audiosciencereview.com/forum/index.php?threads/e14-1-1-600-600-cheap-transformers-some-measurements.10438/

If I breadboard a circuit like the following, could I test it by running a tone generator from my phone at various frequencies into the input and comparing RMS AC voltage at points A and B with a multimeter?

I suppose the main question is what response do you want to measure?
R5=1k and R6=1k will give something like the manufacturer's response.

However if the final circuit opt's for different R5, R6 values
then that will give a different overall response.

A few changes:
- used about 1Vrms in
- increase output cap to 220uF to keep the drive level constant
- measure test point C and test point B
- Perhaps measure at frequencies 10Hz, 15Hz, 20Hz, 25Hz.
  You can do a few points in between those if you want more accuracy,

Make sure the DMM stays on the same range.  Don't let it auto range.
The response is VB/VC.

Thanks Rob. I set it up with your changes. I am inputting 300mV AC at 440Hz. At point C I get 300mV but at point A I get 166mV. That one 1k resistor, R5, has two different voltages on either side so I think something is set up wrong. Do you know could be causing a signal loss across that resistor?

Thanks,

-k

Quote from: kenjib on October 14, 2023, 05:41:32 PMThanks Rob. I set it up with your changes. I am inputting 300mV AC at 440Hz. At point C I get 300mV but at point A I get 166mV. That one 1k resistor, R5, has two different voltages on either side so I think something is set up wrong. Do you know could be causing a signal loss across that resistor?

With input resistor R5=1k and load resistor R6=1k, point A and point B will be about half the level at point C, with A a little higher than B; labels as per test circuit schematic.  The voltage drop expected because the input 1k forms a 1/2 voltage divider with the 1k load. In a basic circuit simulation at 440Hz I get: VC = 300mV, VA = 172mV, VB = 127mV.    The 172mV isn't too far off the 166mV.

If R5 was 100 ohm and R6 was 10k then voltages VA and VB will be closer to VC because the divider is only 100 ohm plus the transformer's DC resistance and a 10k load (in parallel with transformer losses).   The motivation for choosing R5 small might not be to reduce the divider loss it can be more about extending the *overall* low frequency response.

In my response to your first post I mentioned we need to define what response we want to measure.

For a final circuit we want VC in and VB out.  The different R5 and R6 values will also give a different overall frequency response (VC to VB).

The results will have base-line attenuation (the divider), depending on R5, R6, then a frequency response and insertion loss on top of that (the transformer characteristic).

In many transformer tests you will also see VA in and VB out given as the frequency response.   This response isn't really useful for a final response because VA will vary with frequency even when VC is kept constant and that variation isn't present in the measured response.   It can affect the low frequencies a lot.

More often than not, there is ambiguity regarding what response is given in the datasheets.  Typically R5 and R6 are chosen to match the transformer impedance spec.  But what isn't clear is if the response is VC in to VB out or VA in to VB out.  (non audio/telecom applications like power electronics tend to use VA in and VB out.)   This ambiguity is more about matching the manufacturer's data since at the end of the day the final circuit depends on VC in and VB out.

You should measure the base-line levels at 1kHz then the lower frequency response at 20Hz etc.  That will give base-line + insertion loss.   My apologies, I forgot to mention taking the 1kHz measurement.

Another thing you can measure is the actual DC resistance of the windings.  Often the measured results can differ from the datasheet, it depends if the datasheet values are nominal or maximum values.  More accurate DC resistance can help with a more accurate spice mode.

So some annoying details here.  If you measure VC, VA, VB then at least you can capture all the response cases, overall and transformer only.   However the overall response only applies to the R5, R6 values in the test set-up, not entirely representative of a circuit with a low R5 value.

Freq.	Input	A	B	C	B/C
22kHz	117mV	61mV	41mV	118mV	0.3475
20kHz	181mV	96mV	66mV	181mV	0.3636
18kHz	186mV	100mV	70mV	186mV	0.3763
16kHz	192mV	104mV	73mV	192mV	0.3802
14kHz	199mV	107mV	76mV	192mV	0.3958
1kHz	299mV	166mV	122mV	299mV	0.4080
100Hz	302mV	160mV	116mV	302mV	0.3841
90Hz	302mV	160mV	115mV	302mV	0.3808
80Hz	302mV	158mV	114mV	302mV	0.3775
70Hz	302mV	157mV	113mV	302mV	0.3742
60Hz	302mV	154mV	111mV	302mV	0.3675
50Hz	302mV	152mV	109mV	302mV	0.3609
40Hz	302mV	147mV	105mV	302mV	0.3477
30Hz	302mV	140mV	99mV	302mV	0.3278
25Hz	302mV	134mV	94mV	302mV	0.3113
20Hz	301mV	126mV	87mV	302mV	0.2881
15Hz	301mV	114mV	77mV	302mV	0.2550
10Hz	299mV	95mV	61mV	301mV	0.2027

Here's a first pass at processing your measurements.

                    Overall response      TX only
normalized gain:  -8.5dB                -2.9dB
-3dB cut-off:      16Hz                  5.2Hz
-1dB cut-off:      32Hz                  10Hz

Corresponding response plot:


Slightly different normalization, not greatly different to previous values.
                    Overall response      TX only
Normalized gain [dB]    -8.37    -2.95
f_3dB [Hz]                    16.61    4.77
f_1dB [Hz]                    32.64    9.37

Corresponding response plot:

---

Simulation based on 3.9H inductance:

                    Overall response      TX only
normalized gain:  -7.4dB                -2.6dB
-3dB cut-off:      24Hz                  (6.1Hz)        ; (estimate from -1dB point)
-1dB cut-off:      47Hz                  12Hz



The measured response produces lower cut-off perhaps due to the fact the inductance is higher than 3.9H.

Another source of error is the gain normalization.

Quote from: kenjib on October 14, 2023, 05:41:32 PMR5, has two different voltages on either side so I think something is set up wrong.

I wrote a long reply and the forum lost it. Here's the image:

I built a reamp box using a Hammond 145R and posted my experiences here:

https://www.freestompboxes.org/viewtopic.php?t=32742

I chose this transformer because I had it in my junk box

You can get a Hammond 145R for ~$25 from Digikey. So not as inexpensive as what you're trying to use but a much better price than some high-end transformers.

This transformer has a rated frequency response of 150Hz to 15kHz but as you can see from the frequency response graph posted with my design, it does much better than this in practice, especially on the low end. This is because it is being used far below the rated power level of 135 mW.

Hard telling not knowing, since I know nothing to very little about transformers.

How about checking out the recycle pile where old stereo amps / receivers end up.
Or the "junk" store where they sell old audio equipment for next to nothing.

Probably way off base ?

Quote from: Phend on October 15, 2023, 03:11:08 PMHard telling not knowing, since I know nothing to very little about transformers.

How about checking out the recycle pile where old stereo amps / receivers end up.
Or the "junk" store where they sell old audio equipment for next to nothing.

Probably way off base ?

Maybe not ideal.  You would need coupling transformers or push-pull output transformers.  The class-A/single Ended output transformers usually have gapped cores which will make the low-frequency response poorer than ungapped transformers. Some of the transformers out of old telecoms equipment, usually the larger ones, are pretty good but the smaller transformers out of modems have limited low end.

QuoteThis transformer has a rated frequency response of 150Hz to 15kHz but as you can see from the frequency response graph posted with my design, it does much better than this in practice, especially on the low end. This is because it is being used far below the rated power level of 135 mW.

The lower operating power doesn't *cause* a better low frequency response.   The transformer's natural response is probably much better than the spec due to the core size - maybe EI26.   The 135mW implies a certain maximum voltage before saturation/distortion at 150Hz.  At 20Hz that voltage will be lower (and so will the output power).

---

I won't have much time to play with this for a while.  Some quick back of the envelope estimates for the Hammond 145R are

Overall response with low impedance source:
 -3dB 6Hz, -1dB @ 12Hz

Overall response with matched source and load impedance
-3dB 60Hz, -1dB @ 120Hz

So maybe the Hammond 150Hz spec is for the overall response with matched source and load impedance.
There's quite a few assumptions under the hood on these figures.