5F1 OT primary impedance: 8K vs 4K speaker

[King Fan]

I've been reading a lot about primary impedance choices for a 5F1 Champ OT, with candidates running from 5K (supplied in a number of kits) to 7 or 8K (supplied in other kits) up to 17K (as was seen in the reverse-engineered "tiny" 5F1 OT built by Classictone -- RIP). In many posts, speaker impedance wasn't specified, but I assume most folks were talking about a standard 4ohm/3.2ohm "Champ" speaker.

If you're curious, a few of the threads I've been reading:

https://www.tdpri.com/threads/champ-ot-5k-or-8k-primary-has-it-ever-been-answered.1049188/
https://audiokarma.org/forums/index.php?threads/fender-champ-amp-question-8k-vs-5k-pi.870938/

Now the dumb question: *IF* the primary impedance choices were 5, 8, or 17 ohms for ~4ohm speakers, what would be the equivalent primary impedances for an 8ohm speaker? Somewhere around "turns ratio", squaring terms, and "reflected load", my little brain gets saturated and starts to overheat.

 

[NTC]

Multiply by 2. 10 k, 16k, 34k.

 

[YellowBoots]

Pretty sure the desired primary impedance is the same. Adjust the OT to present the same primary impedance to the power tube plate.

 

[Lowerleftcoast]

*IF* the primary impedance choices were 5, 8, or 17 ohms for ~4ohm speakers, what would be the equivalent primary impedances for an 8ohm speaker? Somewhere around "turns ratio", squaring terms, and "reflected load", my little brain gets saturated and starts to overheat.

Assumption: Let's say you can trust the info given by the OT winder.

They usually just give the reflected Primary impedance for a given load. Example: 8.4k primary for a 4 Ohm load.

We can find the impedance ratio by dividing 8400/4 = 2100/1 or 2100:1

To find the reflected Primary impedance for any other load, we multiply the load with 2100. Example for an 8 Ohm load:

8(2100) = 16800 or 16.8k Primary for an 8 Ohm load.

Hmm, lets say I want a 5E7 Bandmaster 3 speaker setup on my Deluxe Reverb. Three 8 Ohm speakers in parallel makes a 2.67 Ohm load.
The impedance ratio of this transformer 2100:1
2.67(2100) = 5607 or 5.6k primary impedance when paired with a 2.67 Ohm load.

Edited for the mistake mentioned by @NTC in post #5

 

[NTC]

Hmm, lets say I want a 5E7 Bandmaster 3 speaker setup on my Deluxe Reverb. Three 8 Ohm speakers in series makes a 2.67 Ohm load.
The impedance ratio of this transformer 2100:1
2.67(2100) = 5607 or 5.6k primary impedance when paired with a 2.67 Ohm load.

Parallel, not series. If they were series connected, it would present 24 ohms (I am sure @Lowerleftcoast knows this. The transformermath is a very good explanation).

Also, isn't a Deluxe Reverb 6.6k:8? The reflected impedance would be 2.475K.

 

[King Fan]

Multiply by 2. 10 k, 16k, 34k.

Pretty sure the desired primary impedance is the same. Adjust the OT to present the same primary impedance to the power tube plate.

They usually just give the reflected Primary impedance for a given load. Example: 8.4k primary for a 4 Ohm load.

I suspect you guys are all saying the same thing. Heck, even hope so. Let me puzzle through this and see if I get close.

Starting with what @Lowerleftcoast says about how mfr's usually label primaries, an "8K" primary on a traditional, single-secondary Champ OT means 8K into 4 ohms, which would be the same as 16K into 8 ohms? That seems to agree nicely with @NTC's note to multiply by 2. And it might agree with the note from @YellowBoots that the "desired primary impedance is the same," if that means you'd want the 8ohms OT to be labeled 8K:8ohms.

I may have that all wrong. But if I have it sorta right, how do I deal with a multi-tap OT, say one with 8K into say 4ohms and 8ohms?

 

[Lowerleftcoast]

an "8K" primary on a traditional, single-secondary Champ OT means 8K into 4 ohms, which would be the same as 16K into 8 ohms?

Correct.

But if I have it sorta right, how do I deal with a multi-tap OT, say one with 8K into say 4ohms and 8ohms?

Just find the Impedance Ratio for each tap.
I am trying to find one of those *charts* for a universal OT where you choose different taps for many different scenarios.

 

[Pete Farrington]

Key aspects of OT design are winding ratio and primary inductance.
The primary inductance provides the foundation that facilitates the reflected impedance on the primary from the loading on the secondary.
So '8K into 4 ohms, which would be the same as 16K into 8 ohms' isn't quite correct, as the primary inductance is only sufficient to support the intended -3dB low frequency for 8k.
With 16k reflected, the -3dB low frequency will be shifted up an octave.

 

[bebopbrain]

8K into 4 ohms, which would be the same as 16K into 8 ohms?

Yes*.


* There is more to transformers than ratios. Once you decide the turns ratio (and core) you need to decide how many turns. You pick a 3dB loss frequency for bass, maybe 50 Hz. When the impedance of the primary (at your bass frequency equals the impedance of the tubes, stop winding. The tubes and the primary form a voltage divider and you will have lost half the voltage. So 16K into 8 ohms will have lost an octave of bass compared to the same transformer used 8K into 4 ohms. For big transformers in, say, the Fender Twin, losing an octave of bass will not be noticed.

OK, @Pete Farrington beat me to it.

 

[Lowerleftcoast]

The full coil is 1 to 6. There are several taps on the coil (2 through 5). Any of the taps can be used to find a desired impedance ratio. Can you find the ratio for a 16 ohm load? As @Pete Farrington mentioned there will be different inductance's for each amount of turns (wire size, diameter, layers, length, iron core, etc will determine the inductance (measured in Henrys)). Efficiency of the OT will be affected by the choice. I once read a piece from the Soursound cat. Basically he mentioned these universal OT's were less desired because only part of the coil was used, making low Henry numbers. Imo, It might make a very good OT if only 1 and 6 terminals were chosen.

(All things being equal more turns, right sizing the coil, and good iron will make the most Henrys. I like to see manufacturers post the inductance numbers. Some of the expensive companies don't even post the primary impedance, smh).

 

[YellowBoots]

@King Fan, I thought you were asking whether the typical “preferred” primary impedances of 5K-17K that you’ve seen bandied about would change if you designed a Champ for an 8-ohm speaker. My point is that the desired primary impedance is dependent on the power tube, not the speaker. The actual primary impedance will result from the OT and speaker you select naturally, but that doesn’t change the fact that a SE 6V6 can run anywhere from 5K-17K—regardless of impedance of speaker.

 

[YellowBoots]

8K into 4 ohms, which would be the same as 16K into 8 ohms

It would be the same winding ratio for the OT, but that’s about it. Headroom would be different, screen current would be different, peak voltage and peak current would be different, and most importantly… it will sound different. So no, 8K:4 is not the same as 16K:8.

 

[King Fan]

OK, sorry guys, just getting to this after a great day outdoors and dinner with friends tonight. I've got a volunteer gig in the AM, but I promise I'll read through all these helpful posts tomorrow and try to wrap my head around it -- or vice versa. Meanwhile, I truly appreciate it.

 

[Pete Farrington]

Once you decide the turns ratio (and core) you need to decide how many turns. You pick a 3dB loss frequency for bass, maybe 50 Hz.

That's not quite how I understand things
To expand on

The primary inductance provides the foundation that facilitates the reflected impedance on the primary from the loading on the secondary.

over most of the audio band, the inductive reactance of the primary will be much higher than the load impedance reflected back from the secondary to the primary.
They're effectively in parallel, but the much higher inductive reactance means that it can be ignored, and the approximation that the 'primary impedance = reflected primary imedance = winding ratio squared x secondary load' works fine.
However, as frequency is reduced, eventually the point at which the inductive reactance = reflected primary imedance.
As the convention that the load and supply impedances are resistive is used, hence the inductive reactance will be 45° phase shifted with respect to the resistive reflected impedance, so that will be the design -3dB point.
Same as the -3dB corner frequency calculation of a RC high or low pass filter.

When the impedance of the primary (at your bass frequency equals the impedance of the tubes, stop winding. The tubes and the primary form a voltage divider and you will have lost half the voltage

See https://en.m.wikipedia.org/wiki/Maximum_power_transfer_theorem

The maximum power transfer theorem works for output triodes and radio / transmission lines, but it's incredibly wasteful and finds little application elsewhere.

The output pentodes we use have a high anode resistance, much higher than the 'through the Vg1-k=0 plot knee' loadline that will provide maximum power output.

Rather than being a voltage divider, a SE output anode uses the inductive energy stored in the OT's magnetic circuit to swing from the saturation voltage V sat to ((HT x 2) - V sat.
See http://www.valvewizard.co.uk/se.html
In the 4k loadline Merlin uses for his example, V sat looks to be about 10V, so he seems to approximate it to be 'close enough 0 as to make no difference')

 

[King Fan]

Whew, thanks, brain trust. I said at the start I didn't understand the variables involved, and your patient explanations really help.

I thought you were asking whether the typical “preferred” primary impedances of 5K-17K that you’ve seen bandied about would change if you designed a Champ for an 8-ohm speaker.

You're right, that was my basic question: If I liked 8K primaries with my 4-ohm speaker, what primary would I want if I used an 8-ohm speaker instead? Your insight that it's actually about the output tube is very helpful.

As an EE tyro, I often need to read, re-read, and re-re-read some basic sources. I'm reminded now that both Merlin and Aiken tackle the OT.

http://www.valvewizard.co.uk/se.html

https://www.aikenamps.com/index.php/output-transformers-explained

Merlin steps through a full SE example using the EL34. I'm lazy enough to wish he'd used a 6V6 instead, but if I re-read enough times, I usually start to get there.

Aiken is also quite lucid on this topic; he walks us through the reflected load idea on impedance, then explains the other variables like primary inductance, wire size and windings, power handling, and mechanical features.

 

[King Fan]

For your amusement, I tried drawing two loadlines in vtadiy. Honestly, I'm just guessing which boxes to fill in and what numbers to use; these are supposed to be loosely 5F1. I marked the 0 Vg line, if that helps, but the locations (and numbers in the label) are arbitrary.

If I did anything right, what do the lines tell us? Are there values (maybe Iq?) I should tweak?

 

[Pete Farrington]

I'm just guessing which boxes to fill in and what numbers to use; these are supposed to be loosely 5F1

The key boxes are V+ and screen grid voltage (these need the cathode voltage subtacting), quiescent current and load impedance.
From that, I guess it should about 280V for the screen grid?

I marked the 0 Vg line

Well done, but it's not super important for class A.

what do the lines tell us?

The 5k loadline looks a bit too steep for centre bias, and is over the max dissipation dotted line.
After having a tinker, with that 350ishV HT, a loadline up around 10k seems to look better to me.
Unfortunately the calculator isn't great for class A / SE, as it requires the user to make an educated guess where the loadline crosses the x axis.
And the power output numbers don't seem to make sense to me.

 

[2L man]

A-class power come low when HV is too high. Input numeric headroom which stop to g1=0V and you see distortion estimates and mathematical power what this swing produce.

Typically Champ produce quite a lot of power in AB-class and 5Y3 rectifiied HV sags. There are at least two mechanisms: OT core saturation for full power low freauencys. Screen current can also increase when frequency is in loudspeaker resonance peak window when load drop drasticly and loadline pass g1=0V well below the knee.

Test 250V anode and 45mA bias and A-class operation come about optimum.

 

[YellowBoots]

Yes*.


* There is more to transformers than ratios. Once you decide the turns ratio (and core) you need to decide how many turns. You pick a 3dB loss frequency for bass, maybe 50 Hz. When the impedance of the primary (at your bass frequency equals the impedance of the tubes, stop winding. The tubes and the primary form a voltage divider and you will have lost half the voltage. So 16K into 8 ohms will have lost an octave of bass compared to the same transformer used 8K into 4 ohms. For big transformers in, say, the Fender Twin, losing an octave of bass will not be noticed.

OK, @Pete Farrington beat me to it.

I was looking at Hammond’s replacement OT for the Bronco (1760C). It has an 8K winding with a primary inductance of 23H. The frequency response is pretty flat right off the left side of the chart. Would you expect the low end response of this OT to remain quite good even if we purposely doubled the nominal speaker impedance? (Eg used a 16R on the 8R tap.)

 

[bebopbrain]

8K winding with a primary inductance of 23H. The frequency response is pretty flat right off the left side of the chart. Would you expect the low end response of this OT to remain quite good even if we purposely doubled the nominal speaker impedance? (Eg used a 16R on the 8R tap.)

The formula for primary impedance is Z = 2 * pi * f * L
In your example the primary impedance at 80 hz is : 2 x π x 80 × 23 = 11.5 k ohms.
This is reasonable for an 8k output stage at guitar frequencies.

I don't think changing the reflected load (changing the speaker) affects bass response.