• 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.

Output Transformer - Silicon Steel or Amorphous?

True, M core plays no POWER function above 500 Hz, it still controls rise time of the planar mag fields being created within the window boundaries. This does limit high frequencies to about 50 kHz with power core stacking and 90k Hz with passive demagnetization stacking.

48% nickle is limited to 3.5kHz for POWER but controls fields well beyond that. 80% nickle works for POWER out to 10kHz. Amorphous core suitable for audio works to 28kHz for POWER .

The only reason to use M core is that above 500 Hz you are only relying on antenna event winding to winding coupling. This means you can design for low distributed capacitance in the windings and coupling capacitance an order of magnitude greater and still obtain 50 kHz. The reason for needing this is to utilize the E Field rather than try to eliminate it. Provides roughly double the information content for small signals coherence. Things like gradient structure in tones and transients, room space seethe, reflections etc.

As you go up in permeability of core you cannot utilize as many sectors in winding, so your distributed capacitance becomes a problem and your coupling capacitance drops. Having said that nickle core is still best for line level and lower signal transform.

Bud
 
Some old curves i found in my VAC (vacuumschmeltze) book.
All depends on the thickness of the laminations. it's NOT that nickel is better at higher frequencies then normal iron.
btwb vacoperm 70 is the name for 80% nickel

An externally hosted image should be here but it was not working when we last tested it.

An externally hosted image should be here but it was not working when we last tested it.
 
Last edited:
I usually establish minimum inductance at about 300-to-500 Gauss AC induction for OT design. It depends on the actual design and application. In some cases you can achieve pretty good specs (typically with lower impedance devices) and in others you have to accept some limitations.
This will affect the transformer size, turns etc. More or less it usually corresponds to Pout around the 0.1 W mark. The performance at low frequency will not change significantly as function of Pout because if you follow this guideline you will get quite high minimum inductance so that all major changes will affect the performance below 20-25 Hz......
For example, if you want to have a good 300B SE OT with best grade M6 able to cope with 13W RMS at 30Hz with very low distortion (i.e. XL/Req = 8 or better) you need at least a 40x50 mm core size! Using this size the max DC (at 75-85mA plate current with "correct" gap) + AC induction at 30Hz will be lower than 9000 Gauss (about half is DC) at full power (turn ratio so that primary impedance is about 4-4.5K for 8R secondary load). That's quite bigger than typical 300B OT's (that are apparently specified for more power.....) you see around. So on this core if you have about 2700-2750 primary turns you will get at least 26-27H for 500G AC induction only! If you want higher minimum inductance, without changing the core type, you need to increase the size because increasing the turns on smaller cores will not improve overall performance as you will be compromising other important things as max permissible induction and insertion loss.....
Making 9 sections (5 primaries and 4 secondaries) where outer primaries have 1/2 number of turns will result in pretty ideal sectioning and a good balance between leakage inductance and stray capacitances. In fact I got about 50KHz at -3dB and very smooth roll-off with no peaks and deeps.

What do you guys think about these OPTs from Novosibrisk (NEM)?
This 60 kg monster mostly made from M4 steel has output transformers with 50x50 mm core size for 7 watt output power (5K primary).
 

Attachments

  • _IGP4143.JPG
    _IGP4143.JPG
    520.6 KB · Views: 690
Some misinformation in this thread, particularly regarding contribution of silicon steel core at high frequencies, and capacitive coupling between windings.

Iron core is NOT "out of the picture" above 500 Hz. Permeability of silicon steel at 10 kHz is ~50% of initial, and at 20 kHz ~30%. An iron core is completely out of the picture only at about 100 kHz.

For this reason, 1) keeping the winding close to the iron core (like in toroid transformers) is an effective way of reducing leakage inductance; 2) small size transformers may have low leakage inductance without interleaving.

Electrostatic (capacitive) coupling is bad in audio transformers: it introduces a low quality capacitor in the signal path. For this reason, high degrees of interleaving, advertised by several manufacturers as unconditionally good thing in a transformer, has a serious drawback. Good transformers use Faraday shield to eliminate capacitive coupling.
 
Last edited:
... Having said that nickel core is still best for line level and lower signal transform. Bud

I was just reading this older post by Bud. I have a few OPTs including amorphous, and the one I'm using in my system is one of Bud's earlier commercial SE models. Sounds very sweet and natural. Fortunately I have one more SE of his as well, plus a pair of his PP. I respect his thinking.

I'm trying to do as above - work out the best option for a plate choke for the input stage. Dave Slagle for one can build with nickel and that's an option I'm interested in. Probably with a 10Y valve so needs 200H inductance.
 
Nickel alloys used in transformer cores have extremely unpredictable permeability. Other than well-known changes imparted by conditions of annealing, permeability of processed Ni cores is dramatically and irreversibly affected by temperature, age, previous history of magnetization, mechanical impact, location of planets in the sky, etc, etc.

I tested a few samples including EI cores from antique audio transformers and chokes, cores from 1950s-60s (UTC, Triad, etc), modem transformer cores, and specialty strip-wound encapsulated toroids. 80% Ni cores were all over the place, with some extremely high, yet other (particularly older ones) lower than non-oriented silicon steel. One lot of Midcom 671-8422 transformers had unbelievable 480 H inductance per side. This is in a transformer 3/4 by 3/4 inch and winding DCR of 120 Ohms! I thought there was something wrong with measurement, but in the same test larger Midcom transformers (671-8000) were 11 H per side, as of specs.

Bottom line, Ni core transformers may have low distortion and great characteristics when new, but only God knows how long the happiness will last. Silicon steel is more predictable, and, although its characteristics change with time, the change is relatively small. From my experience, aging of of silicon steel mellows audio transformers and makes them more refined.
 
Why?
10/801 working with 7-10k OPT.
10k at 20Hz is only 80H.

You made a good point, so I tested it out. 10Y with a single 126C at 100H against 2 in series at 200H, used just as plate chokes.

Conclusion is that the 200H of two in series is clearly better sounding - more detail and clarity, better timbre, more treble definition and air. Better all round in fact. The single 126C clearly works, but it's rather flat and dead sounding in comparison, and with an audible loss of clarity. These are cheap units and anyone thinking of using them with higher Ra tubes like 26 and 10Y should, in my opinion, go for two in series. If I order a different plate choke for the 10Y I'd be wanting 200H on the basis of this test, but with a different make who knows - results might not be the same?
 
Andy, that's a great finding...I would never thought about this like this...I would have expected the opposite: Better bass response with more inductance, but a less engaging sound as the dc-resistance goes up like a cheap choke vs. an expensive one...

So, great finding, will try this
 
more detail and clarity, better timbre, more treble definition and air

If one thinks about a plate choke as a current source load, these are exactly the results that I have experienced with active CCS loads when replacing an inferior current source with one that is superior. For example, a cascode DN2540/DN2540 vs. a single 10M40 or an IXYS 8N50D2/1N100D vs. a DN2540/DN2540.
 
I'm not saying it's the best. It is better than the DN2540 cascode. There are caveats: if the current is less than 10mA you might want to look elsewhere Although I am currently using it as a plate load for a 6SN7 @8mA and it sounds very good), also the IXYS cascode really needs ~30V across it to perform well (also allow for the anticipated signal swing). If you want to experiment, Kevin Carter @K&K Audio sells CCS kits that are very handy and real cheap and they are supplied with IXYS MOSFETs.
 
Two audio chokes in series halves the winding capacitance, extending high frequency response of the combo.

Even better idea is two dissimilar chokes in series, a cheap LF choke with high inductance and high winding capacitance, and good quality HF choke with low inductance and low winding capacitance. The HF choke is something like 1-2 H with a highly sectioned winding and gapped ferrite core.

LF choke provides for low bass extension by virtue of its high inductance. HF choke determines treble extension by virtue of its high self-resonance frequency.
 
One lot of Midcom 671-8422 transformers had unbelievable 480 H inductance per side. This is in a transformer 3/4 by 3/4 inch and winding DCR of 120 Ohms! I thought there was something wrong with measurement, but in the same test larger Midcom transformers (671-8000) were 11 H per side, as of specs.

That is nothing special by todays modern material standards.
I measured a just finished autoformer volume control, wound on a nanocrystalline toroidal core with Afe of 0,86 cm² (some 0,13 inch²).
DCR 31 ohm; inductance 160 H.
Enough winding space on that toroid for about 4 times the turns to get near 120 ohms DCR; that would give an inductance of some 2400 H :D
 
autoformer volume control, wound on a nanocrystalline toroidal core

Yes, these nanocrystalline cores are amazing! I am making a similar TVC with a toroid Vitroperm core from Vacuumschmelze (sold by Mouser). Low number of turns for high inductance, so an easy DIY project.

However, how stable these amorphous/nanocrystalline materials are remains an open question. High magnetic properties depend on crystal structure, and all materials re-crystallize slowly even at room temperature. I noticed that there are two types of Vitroperm cores, one encased in rigid plastic sheath, and the other (cheaper one) coated with epoxy. Although core material is the same, the encased core has 3 times the guaranteed permeability of the epoxy-coated core. This means one thing: permeability is extremely sensitive to any kind of mechanical strain, exactly like in 80% Ni cores.