• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Audio Note M10 Clone

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I have bought a CNC winding machine, and try to use it to copy the AN M10 output transformer
IMG_0454.JPG
 
I made one, (a hand winding one) back in 1960. I will post a photo when next back in my workshop. It uses a counter from a pre 1960s gas meter on a short piece of bowden cable to the winder.

If you unwind the transformer, measure the wire gauge and count the turns, that is all you need to know. Laminations are widely available and off you go.
Happy days.
 
First of all is confirm the turn ratio. M10(old version) used Tran027 for the output transformer and the turn ration is 33:1. But I doubt about it. Two reason:1, if it is 33:1 is correct, second impedance is 600R, the prime impedance will be 600*(33:1)^2=650K?;2, the output stage is 5687 parallel, the output impedance will less than 6.4K(the typical operation condition of 5687 )

IMG_0456.GIF
 
Ex-Moderator
Joined 2011
Found this on the web:

'Now to some comments about the measurements, may I say that Andy Grove and I will be in HK on the 18th May, so perhaps we can meet and talk during one of the seminars that Elephant will hold?

A.)
Could I first point out that the transformers you have are not M10 transformers. The M10 transformer uses a high permeability nickel iron C-core and silver windings, the core makes a huge difference to the transformer, not only sonically, but electrically as well. The material has much lower losses and far, far higher permeability, and because of the C-Core shape it’s easier to make a high tolerance gap in the magnetic path, which is important for such a transformer.

B.)
The transformers you have are design Trans-028, which is the new design, we don’t use the 027 anymore. The 027 was approximately 33:1 with a CT. The original M3 used a switching arrangement on the rear to ensure that in both balanced and unbalanced output modes the entire secondary was used, to ensure that the entire length of the secondary wire passes through the primary magnetic field. The new transformer 028 has a ratio of approximately 18:1 balanced and 36:1 unbalanced, this is to give a higher maximum voltage output, and more gain at the balanced output, so that the preamps will work well with low gain, transformer input power amps such as the new Gaku-On. The problem of the secondary wire passing through the primary field was solved by using a bifilar wound secondary. This also means that the AC impedance to ground from the + and of the secondary is very well balanced, which is important for true balanced operation. The 028 also includes a primary to secondary screen, this isolates the secondary winding, which further improves the performance of the transformer as regards to hum when used either in balanced or pseudo balanced operation.

C.)
The 028 was primarily designed to work with a 5687 operating at approximately 8 to 10mA DC quiescent current. This current will bring the core material out of it’s initial permeability area, this is particularly important with the lower grades of silicon steel, such as those normally used in laminations. The use of laminations is also problematic because the field has to pass across the grain structure at various angles other than in the preferential direction, at angles other than the preferential direction the material properties are quite poor in comparison. If you excite the primary with only a volt or so of AC then your results will not be representative, the material will be operating only in it’s initial permeability region.

So, really to test the transformer you need to pass 8mA DC through the primary winding and/or excite the primary winding with at least 10Vrms.

To look at it another way, would you apply 10V and 8mA DC to a moving coil stepup?

To do that would be to operate the transformer in an alien environment, and the results you recorded from such an experiment would be inaccurate.

If the gap in the transformer is much larger then its reluctance becomes predominant, so larger SE output transformers will give some kind of sensible results when measured this way. The gap in the 028 is not large and the material is a larger part of the magnetic circuit, so excitation is far more critical.

Another problem with laminations is that it’s quite difficult to create an accurate gap, although here we use a special clamp so that we can squeeze the lams together during assembly, but it’s possible that the gap has moved since assembly, this can happen during shipping, especially when ship by air where it is subjected to minus 50-60 degrees minus, or sometimes the varnish we use to stop the laminations rusting penetrates into the gap, causes the paper to swell and opens the gap up slightly, we do final test the transformers before they leave the factory, but there is still a small possibility, the gap hs changed after that although I really doubt this is the case.

D.)
The 028 is designed to offer an extremely high impedance load to the output valve, this is one of the secrets behind the M3 and higher end AN preamps, like the M10, the transformer is primarily intended to work into an open circuit, or at least a load of around 5k to 10k, similar to that which the balanced input of an amp such as the new Gaku-On presents, the reflected load would be 1.6MOhm. But even a load of 600R results in a reflected load of 194k, which is somewhat higher than one would be used to when working with a regular output transformer, where the reflected load from 8R is usually well below 10k, and usually only a few kilohms. If you can imagine the difficulty of designing a 211 output transformer with 10k, or as we use even higher than that (20k/8R), then designing a transformer whose primary impedance will be almost infinite is a feat unto itself.

E.)
You didn’t specify the source and load you used for your tests, so I assume that you drove the transformer direct from your analyser. The source impedance of a 5687 when used in a M3 or similar preamp is around 5k or 6k, so the most important factor in the transformer design is the primary self capacitance. Leakage inductance is a secondary consideration. If you drive the transformer with a 600R source such as the output of your Panasonic machine (BTW: I have one like it at home!) then it is like smashing the primary with a hammer, remember that even with 600R on the secondary there is very little damping of the primary condary leakage inductance, this damping comes from the series resistance of the source. Essentially it’s like driving an LC filter with a mismatched source impedance; it will ring.

The HF frequency transient response is a compromise; there will always be some ringing. Remember that this transformer has a more or less primary impedance of infinity, and is driven by a source impedance of 6k, that’s a little bit different to a regular output transformer. That ringing can be damped somewhat by loading the secondary, you can prove this to yourself by varying the load resistance, or even as “another well respected transformer company” do, use a Zobel RC network to smother the transformer secondary at high frequencies and cover up ringing. This however does not have a good effect on sound quality in my opinion.

You didn’t specify the frequency where you measured distortion, this is an important factor, as the AC flux density varies with the inverse proportion to frequency, also, when measuring transformers without the aid of an FFT analysis, it’s very easy to read the a THD measurement and see magnetically induced mains hum. But overall silicon steel laminations do produce distortion, there are several components of this distortion, and it differs whether the transformer has a DC bias field, and where it is operating on its BH curve. Laminations, because of the field crossing at an angle to the grain, produce more than a C-Core would, but even then, the nickel irons are far superior in this respect, better than anything I’ve tested (with my proprietary B/H curve tracer, built with valves), even the later amorphous materials which exhibit an odd memory effect not present in soft nickel irons.

But, really, there are many more things to worry about than having neat rectangular waves on the scope. Some years ago I did a series of experiments with a new type of topology of solid state amplifiers (my design). This topology allowed me to achieve slew rates of over 1000V/us, without slew limiting, and distortion figures so low that an AP System 2 could measure virtually nothing in terms of distortion, less than 0.0005%. The amps were also unconditionally stable, you could put a 1uf capacitor directly across the output and drive a 10kHz full power square wave into it with no effect. So, as far as measured parameters were concerned they were perfect, better than any solid state high end gear I’ve ever heard of, but the sound?'
 
A paralleled 5687 is pretty flexible and would work OK with a lower step-down ratio. 15K:600 is a super common transformer that will still present a very high load to the 5687's. Heck, if you choke load those 5687's and use parallel feed you could really have some options for the OPT.
 
I made one, (a hand winding one) back in 1960. I will post a photo when next back in my workshop. It uses a counter from a pre 1960s gas meter on a short piece of bowden cable to the winder.

If you unwind the transformer, measure the wire gauge and count the turns, that is all you need to know. Laminations are widely available and off you go.
Happy days.



Thank you, it is really helpful if you can post some photo of yours. I haven't winded it yet and will measure the inductance of the core by wire a test coil
 
Ex-Moderator
Joined 2011
seems more realistic to me....or a 5:1 ratio....

Thanks for post this email for me, I have read it before, peter is mainly talking about Tran028(the new version) for M10, but can't find the schematic of new M10

Peter Qvortrup clearly stated "The 027 was approximately 33:1 with a CT." and the reason why they switched to the 028 - to get higher impedance, but otherwise I would guess the schematic stayed pretty much the same.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.