• 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.

CCS design for OPT low frequency measurements

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'm currently digging into distortion measurements of OPTs. For proper SE transformer measurements, I need a DC bias with a high impedance, so a current source will be ideal.

Assuming the fact it needs to be able to dissipate quite some power and it will be used to measure the LF distortion and core saturation only, are there any hidden waterstones to watch for, such as a potential self oscillation? Do you have experience with such devices and what are your opinions and recommendations?

Many thanks!
 

PRR

Member
Joined 2003
Paid Member
...distortion measurements of OPTs. ...I need a DC bias with a high impedance...

A pentode?

If you need to distort the iron, the "CCS" voltage compliance must be greater than the way the OT will be worked in the final design. On the face of it this means a bigger pentode and power supply than what you intend to finally build. Generally build a 30W amp, map-out OT limits, then build a 10W amp.

Over the fence: you get the same results abusing the secondary. Voltage is 30X less and current is 30X more. This *may* happen to fit some available sand-state parts and heat-fins and *may* cost less.
 
This is another though - build a universal tube stage for transformer measurements. You can add a lot of local stage NFB before the transformer, to linearise the stages and reduce their output impedance if needed. You can always add resistance in series to simulate higher Rp tubes.
 
Hey Elvee, many thanks for sharing! It seems you've spent a lot a work on this project.
Yes, quite some time, but the CCS "engine" itself is rather simplistic

Your idea together with PRR's suggestion of reverse measuring the transformer might do the trick?
Reverse measurement could ease matters, but if what interest you is purely saturation, that is not even necessary, you just need to pass the required current.

Even for full characterization, MOSfets with Vds max >1000V are now cheap and easily available
 

45

Member
Joined 2008
I have never found any usefulness in measuring saturation especially because what would one define as saturation from the point of view of the amplifier? The technical definition is not useful at all because distortion at low frequency will be excessive well before the core has really saturated. If I get the desired inductance and distortion @40 Hz with small and medium voltage (typically corresponding to 50 and 2000 Gauss AC) then the I know if it will work as designed or not.
 
45,

The main purpose is to measure distortion, which increases with magnetizing current. I don't mean saturating to the extreme, which IMHO is yes, useless.

Elvee,
I also had the idea about 1kV MOSFETs in mind. They're not horribly expensive and.. I've got a huge stash of heatsinks. :D
 
45,

The main purpose is to measure distortion, which increases with magnetizing current. I don't mean saturating to the extreme, which IMHO is yes, useless.

Distortion coming from magnetizing current only becomes relevant below 25-30Hz and is application dependent.
Moreover the magnetizing current distortion has almost ZERO impact on the amplifier IMD because there is no or very weak frequency correlation.....
Magnetizing current distortion is the least of all possible evils if the transformer is done right which I think you already do. If you want to make truly irrelevant then design for max 1T @20Hz also with C cores...
 
Can you elaborate further about other distortion mechanics in transformers that would bother you in the first place? IMHO, I know from experience that sectioning matters a lot. I don't have objective data, but subjectively I've heard horrible musical performance from uneven sections, such as SS-P-S-PPP for example, instead of S-PP-SS-PP-S, which should be the correct interleaving way. The first example might have some hidden gold for guitar effects. :D

By default, I do design my transformers at 1T &20Hz. Sometimes I go up to 1.2T with GOSS.
 
You could, for example, apply some DC voltage from any power supply thru an 8 ohm (or any convenient value) to the secondary winding while measuring frequency response, distortion or whatever. The applied voltage and the resistor determine the current in the secondary. Set it to the current you would have applied into the primary multiplied by the turn ratio.

The resistor may become hot, but it's easy to manage !

Yves.
 
Can you elaborate further about other distortion mechanics in transformers that would bother you in the first place? IMHO, I know from experience that sectioning matters a lot. I don't have objective data, but subjectively I've heard horrible musical performance from uneven sections, such as SS-P-S-PPP for example, instead of S-PP-SS-PP-S, which should be the correct interleaving way. The first example might have some hidden gold for guitar effects. :D By default, I do design my transformers at 1T &20Hz. Sometimes I go up to 1.2T with GOSS.

I am not aware of interleaving and sectioning causing troubles at low frequency (i.e. below 50-60Hz) in audio transformers. It might be a problem thinking about distribution of mmf but really never looked into it. That has surely to do with leakage inductance so it's higher frequency stuff. It might be a problem for measurement transformers like those used in geomagnetism.

The principal sources of distortion of a well made transformer at low frequency are the source resistance (which includes the copper resistance as well), saturation and its limited inductance (reactive load). Some distortion is due the material type but is never really big enough to cause troubles. Even in the case of a PP transformer with very high inductance the source resistance of tubes will never be low enough to go below some threshold which is also signal dependent and material dependent as well.

The dependence on the material is of two types: at very low level is strictly dependent on the hysteresis loop and the initial permeability (so it is smaller for materials like mumetal, radiometal etc.). This cannot be changed with any design! It's intrinsic. However even for standard grade M6 steel we are talking about of 0.5% at worst at 20Hz and it drops very quickly with frequency so that at 50Hz is already of the order of 0.05%. Radiometal is 10 times better and mumetal is 100 times better than M6.

Increasing the signal it drops as permeability increases and goes down by about 10 times! Should one worry about this? No because the low level where it's max corresponds to few mW output (i.e. AC induction of the order of 10-20 Gauss!!) generally speaking for typical output transformers. At 100-200 Gauss it has already more than halved.....

On the other end if you limit roughly the max induction at about 1/2 of saturation induction (even better if just below 1/2) it won't be worse than that either....if you go above such limit it will start to increase roughly exponentially like hard clipping in amplifiers.

Because of this "funny" behaviour hysteresis distortion as function of frequency doesn't cause relevant IMD at low medium levels. Electronic circuits instead have a fairly precise relationship between THD and resultant IMD and it is due to the fact that electronic non-linearities generally distort audio signals without regard to frequency. So the low level distortion, because the circuit is supposed to be quite linear enough, will not cause any trouble both as IMD products and THD itself (being the level low enough in 99% of cases to be below audibility threshold).

Same order of magnitude of distortion but less difference between the 3 materials with source resistance like that of a 300B. Gets worse if source resistance increases. This kind of distortion at high frequency will be design dependent and in particular will be related to the AC resistance of the windings. To keep it low a max AC current density of 0.8-1 A/mm^2 is enough.

So we are left with how much inductance one needs to get some reasonably low distortion at 20-30Hz. This can be summarised as THD vs 2*pi*f*L/Req ratio (where Req is the parallel combination of tube plate resistance with nominal primary load and copper resistance. See in the RH4). By this one can see that it's more relevant for SE. Already in a favourable case like 300B with a 40H 5K transformer for Bmax=0.8T @ 20Hz it is unlikely it will be less than 1-2% with M6 steel. In less favourable cases it will be more. However because there is very little or no music below 30Hz and loudspeakers generate a lot more distortion at low frequency this isn't a problem at all.....it's just bench/advertisement problem....:D
 
Last edited:
Excellent post! Many thanks!

In summary, the most significant source of distortion is the relation between primary inductance and source resistance. Also, measuring distortion (permebility related) at very low levels could be rewarding, because the microdetails are there. What about the distortion in the HF region due to increasing Cp and/or Ls? In practice, we seldom need purity above 16kHz I guess.
 
Also, measuring distortion (permebility related) at very low levels could be rewarding, because the microdetails are there.
You can't do much about it because if you try to minimize too much then you get in trouble in other areas like headroom for instance. If you want to get less distortion at low level then use high initial permeability materials like nickel alloys. I like a lot the radiometal (or 49% Nickel or 4350 alloy...its name depends on the manufacturer with very slightly different specs) in general because it is possibly the best compromise. It has initial perm of 4000-5000 at least (so its minimum perm is almost equal to the max perm of M6!) but it has 1.6T saturation with very little hysteresis (so very low loss) and is perfectly usable up to 0.8T with low distortion (or 1T with some very acceptable lowish distortion if happening around 20Hz). EI laminations of this material also don't have the issue M6 has because the Fe balance is non-oriented.
Amorphous is not an improvement in this regard, IMHO, because its permeability is not better than M6. I haven't measured it though but I am sure that to get the same inductance with the same core size one usually needs to wind more turns! It only has lower losses.

What about the distortion in the HF region due to increasing Cp and/or Ls? In practice, we seldom need purity above 16kHz I guess.

I am very picky about inter-stage phase PP balance and extended FR but only because I still want the OPT (or the input transformer when I use it) to determine the FR and THD of amp. However in less demanding application I use "less appealing" transformers as well with good results.
But really HF distortion is not a big issue because the power needed is usually 1W or less above 10 KHz. Only few special cases where a few watts are needed. So it's easy to get low distortion at this power level and any kind of IMD will be low as well. Sometimes I have seen high THD even at low power level but in those cases something was really wrong or bad.

P.S.
The only real problem with radiometal is that it's not easy to find and when one finds it only comes is small size lamination that are only suitable for input or small-to medium interstage transformers. But I guess that if one orders a good amount they can do it....
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.