Output transformer impedance

[TurboGuitarMelton]

What effect does having a higher primary impedance have on the tone of an amp?


From my research I have found that output power will be lowered. Will the frequency response change? More even harmonics?


Mainly asking to investigate the odd 40-18110 classictone champ transformer. It has a 17k to 4ohm ratio as opposed to the normal 5k to 4 or 8k to 4.

[shooter]

hi, lo are relative.  what's important is matching the tubes desires with the transformers spec.
picture a bell curve, the very top is the "matched" point anything either side is mis-matched and as the curves so EDIT:show, less than ideal

[mresistor]

Shooter ..  that predisposes that the best "tone" is at the optimum point on the curve..  It could be that better tone could be had at different points or with slight mismatches in impedance.


17K is pretty odd indeed.   Mojo's 5F1 ot is 5K primary . The Hammond 1750C is 5K or 8K primary.  I'll bet the original Fender tweed OT was 5K to 8K as well.

[shooter]

I can't see tone on my scope so it's all in your head 


 

[TurboGuitarMelton]

I contacted classictone and this is what they had to say:



"The difference between our 40-18110 and our 40-18115 is that the 40-18110 breaks up very early.
Although the 17K Ohm primary impedance is an unusually high value, our 40-18110 is an exact copy of an actual 1958 5F1 Champ output "

The Hammond 1750C is 5K or 8K primary.

How does 5k and 8k compare? I understand that its all relative but there has to be some kind of difference in the frequency response.


I was reading on the valvewizard that a higher impedance gives more second order harmonics at the expensive of some voltage (power)


"A higher impedance or lower supply voltage would allow us to bias deeper into class A (i.e., hotter), for more second harmonic distortion. Using a slightly lower impedance would force us to bias colder, which tends towards a more 'raw' overdriven tone."


I had a champ once with the 17k transformer. I seem to remember it sounded really dark (less highs) and very compressed.

[tubeswell]

I need to look at a loadline


Assuming for a second the the 5f1 screen voltage was set to 250V (and I know it isn't, its more like 300, but I'm using a datasheet chart and I can't be effed drawing grid curves for Vg2 = 300) here's a fictional example to describe differences in impedance loads for a 6V6 idling at 340V with 35mA plate current (i.e. 12W plate diss). The bias voltage is 15.5V so this hypothetical example isn't using a 470R cathode resistor (it'd be more like 418R allowing an extra 2mA for screen current)

• with a 5k load, you'd need 15.5 - 5.5 = 10V positive swing at the grid to drive the plate voltage down by 100V 200V
• with an 8k load, you'd need 15.5 - 8.25 = 7.25V positive swing at the grid to drive the plate voltage down by 100V 200V
• with a 17k load, you'd need 15.5 - 10.75 = 4.75V positive swing at the grid to drive the plate voltage down by 100V 200V

(Edit: whoops I just noticed an error on my previous chart!  - where is says '100V swing', that should be 200V swing. The principles are the same - higher load, more gain, more harmonic distortion - and the attachment is fixed  )

So the higher the load, the more gain you get (all other things  - screen voltage and bias voltage - being equal) but you end up clipping the 6V6 grid sooner, so you are wasting pre-amp gain and you end up with a muddier signal. (And the more mush that goes into the OT, the less output power per extra amount of mush put in - i.e. decreasing marginal extra mush)

You also risk getting into highish screen dissipation when you drive the amp harder (the dotted black curves are screen current, and they shoot way up as plate voltage begins to drop below about 30V). However, this really isn't tooo much of a problem with the 17k load line in this case (and in the 5F1, the 10k screen node supply resistor eats up quite a bit of screen current anyway).

Also, you can see the spacings between the grid curves along the 17k load line gets smaller as you move up the load line, compared with the spacings between the grid curves on (say) the 5k load line, which are more even as you move along the line. So as the signal swings get bigger with a 17k load, there is more harmonic distortion (than there is with the more linear 5k load)

As I say, the 5F1 in the Fender schematic is running the 6V6GT with a screen voltage of about 300, so that nudges all the grid curves up a little. Also the bias voltage is 18V through a 470R resistor, so you're looking at 38mA tube current (but about 36mA plate current). So that makes the transconductance a bit higher. But the slope of the loadlines would be the same.

[shooter]

fwiw;
I've went out to X2 on impedance and X.5, the result to a non-guitarist was underwhelming for "sound/tone".  It did "darken" at the margins but the (in my case) KT88 just took it in stride
so that said, mess with speaker load to give you an "effective shift" and listen, no solder or heavy iron lifting required 


I believe impedance is a non-linear, frequency affected, math problem, best solved experimenting, just watch the tube for bad glow-ez n you should be good

[TurboGuitarMelton]

Wow tubeswell!!! Thank you for your response! That was very informative! Do you have any books or websites you recommend I check out so I can learn more about load lines? This kind of information seems like a missing link (of many) in my understanding of designing tube circuits.



Thanks for your reply too shooter! I think in this instance I am splitting hairs but I am a guitarist looking for a very specific feel from the amp. The slight reduction in perceived highs and increase in compression/harmonic distortion is exactly what I'm looking for.


I got another reply from classictone that I think is VERY interesting and contradicts a lot of stuff I've read on other forums:



"The treble response will be about the same between our 40-18110 and 40-18115. With the higher primary impedance and resulting higher primary inductance of our 40-18110, the bass response should be slightly better. But you most likely will not hear it since it overloads so much quicker. After all, we are only talking less than 5W here."


Call me crazy but I am completely obsessed with champ circuits and low power guitar amps. (thats about all i ask about one here) I am a gigging guitarist and have never run into a gig where my trusty little 5F1 clone hasn't been loud enough. I've come close a few times though. Lol A lot of that is due to the fact I play jazz/roots/avant music with some very aware and dynamic-conscious musicians. I figure a champ maxed is about as loud as an acoustic piano, saxophone, cello, or trumpet. It also responds better to small nuances in my experience than a larger cleaner amp.

[shooter]

I like the "little guys" myself, never build anything over ~20W audio power.  play with speakers, if you have a couple cabs and you're normal speaker load is 8 ohm, wire 16, then 4.  Get real crazy and do 8 and 4 ohms for 12 ohm.

[HotBluePlates]

... Do you have any books or websites you recommend I check out so I can learn more about load lines? This kind of information seems like a missing link (of many) in my understanding of designing tube circuits. ...

RCA Application Note 78

Crowhurst Basic Audio Vol 2

Start with those, then ask your questions.    This topic is waaaaaaayy simpler then most make it out to be (including myself, until the basics finally clicked into place).  Make sure you have a death-grip on Ohm's Law, and understand the difference between Resistance, Impedance, and Reactance.

Later, you can look at other books here.

... what's important is matching the tubes desires with the transformers spec. ...

It is damaging to one's ability to understand this load impedance if we think of the tube as "desiring" anything.  Think only that there is some perfect "resistance" that results in the highest "power dissipation" with a set amount of voltage (dictated by the power supply).

The load impedance at the primary dictates how much current can flow, but the current is also limited by what the tube can manage.

Any load above/below the ideal "resistance" that results in the highest power dissipation simply gives less power, more distortion, or both.

... I got another reply from classictone that I think is VERY interesting and contradicts a lot of stuff I've read on other forums:

"... higher primary inductance of our 40-18110, the bass response should be slightly better. ..."...

Classictone's answer is related mostly to the construction of the transformer and it's resulting characteristics.  That's why they mentioned primary inductance resulting in "better bass response" but this is a known factor in transformer performance.

Put simply (and related to the things I mentioned earlier that you should know):

        -  Higher primary inductance = Higher Reactance at some low frequency.

        -  This leads to larger voltage drop for a given current pulled through the primary.

        -  Same Current x Larger Voltage = Larger Power

So less primary inductance means less bass output at some low frequency.  They said "better" but that should be read as "more" not "nicer" (quantitative, not qualitative).

What effect does having a higher primary impedance have on the tone of an amp?

From my research I have found that output power will be lowered. Will the frequency response change? ...

Back to your original question, and what tubeswell pointed out:

Based on the last page of the 6V6 data sheet, distortion does increase.  Perhaps more importantly, the balance of even & odd harmonic distortion is changed (quantity & relative mix changes repeatedly over the range of possible load impedance).

Keep in mind when you look at that graph that push-pull amps cancel even harmonic distortion generated by the output tubes; most/all of the output tube distortion in those amps is odd harmonic.  Even harmonic is less-obvious & darker; odd harmonic is brighter & more obvious in smaller amounts.  Neither is necessarily bad/good.

[TurboGuitarMelton]

Thanks guys! Im excited to get my hands on one of those books and dive deeper.


I got the small 17k transformer in the mail today and swapped it for the "normal" 8k one.


From a guitarist stand point and to my own ears here is what I hear: The amp is definitely darker sounding and much more compressed and mushy. I love it! A lot of the piercing tones I would get at max volume are gone and the amp seems more "vintage" or "smoky". I suspect there is a shift to more even harmonics. That sounds like what im hearing.


The break up is there at lower volumes but its harder to notice because its a smooth kind of break up.


- more of a growl than a crunch

[shooter]

do you have your "old" cathode VDC readings?  If so take a new reading and compare.  "should" give you a "sense" of the tube working harder (lower vdc) or easier (higher vdc)

[TurboGuitarMelton]

do you have your "old" cathode VDC readings?  If so take a new reading and compare.  "should" give you a "sense" of the tube working harder (lower vdc) or easier (higher vdc)

No, I should have taken them!


Here are my current readings with the 17k transformer


Cathode voltage 17.49
Cathode current 37.48ma
Plate voltage 322.4
Screen Voltage 289.9

[shooter]

if TGM halves his speakers, are we back to start with impedance math?  (ASSUMING 17K doubled original)EDIT: static conditions
I'll concede tone is NOT = to math 


get used to documenting, computers have been around since the analog days, paper n pencil a few years B4 that

[HotBluePlates]

Here are my current readings with the 17k transformer

Cathode voltage 17.49
Cathode current 37.48ma
Plate voltage 322.4
Screen Voltage 289.9

Under 12w, so the amp is happy as a clam!

if TGM halves his speakers, are we back to start with impedance math?  (ASSUMING 17K doubled original)EDIT: static conditions ...

Static is the key word...

Classictone doesn't provide DCR values on their spec sheets.  However, Hammond specs their (8kΩ) Champ output transformer at ~255Ω.  It's reasonable to assume Classictone is similar with their 8kΩ Champ OT, and about double that with their 17kΩ OT.

Inductive Reactance = 2πfL ---> if frequency is 0Hz, Reactance falls to 0Ω.  Impedance is the vector-sum of Resistance & Reactance, so with Zero Reactance only the DCR is left.  So No Audio = Primary is like a ~510Ω plate resistor.

510Ω x 37.48mA = 19.1v dropped across the OT primary.  You'd have half that with half the DCR (if screen voltage stays the same, idle bias & current will probably stay the same with either 8kΩ or 17kΩ OT).

You might notice "510Ω" is a small-% of "17kΩ primary impedance."  We want it that way, so output power is not wasted heating up the transformer.

The value of using a transformer in a power output stage is:

• Keeps high voltage d.c. off the speaker
• Steps up the small output tube plate current to a large current needed to move the speaker (turns kΩ into a-few-Ω)
• Passes more of the supply voltage to the plate, increasing the available plate-voltage swing (which increases power output)

Once there is a.c. across the OT primary (due to a.c. signal driving the tube grid & causing plate current-swing), then the primary inductance comes into play and causes that a.c. to see the 17kΩ primary impedance.  Because this load is reactive, and because the transformer passes a.c. power from primary to secondary with relatively little loss, the apparent power "dissipated in the primary impedance" is actually transferred to the secondary and pumps the speaker back & forth.

Small-current, Large-volts = High Impedance
Large-current, Small-volts = Low Impedance

The sooner the connection is made that tubes pass relatively small-current but swing large voltages, and that there is some optimum load "resistance" that allows the tube to "make the most of the available supply voltage" (i.e., generate the most output power), the faster one sees the output stage & how it relates to the OT primary impedance.

Aside from there being some shortcuts based on observation of the tube's characteristics in relation to the size of the load that happens to deliver the most power, we can dismiss the notion of the "tube wants some particular load."

The loadlines below are from a different example I posted elsewhere, so plate voltage is higher & ≠ screen voltage.  However, the basic premise applies: max power happens with a loadline that results in the largest rectangle (plate-volt swing x plate-current swing = power output); higher or lower loads than optimum simply deliver less power (though could bump into some limit of the tube's ratings or safe operating area).

[shooter]

"tube wants some particular load."

I think a light bulb lite
you're right, it's the designer that wants a tube to operate at a given point (tubes happy place:) so that the MOST buyers will be happy.


my hangup is, using speaker load to get to 17k vs buying a transformer to get to 17k, will the tube operate at the same place, since it no's nothing of speaker or OT?
IF so, I will concede it's cheaper to put in a $50 OT than a $150 speaker, but not very practical if you want 2 operating points, ie; i'm gigging at a biker club, go with 2 speakers, tomorrow I do jazz, go with 1 

[pdf64]

...my hangup is, using speaker load to get to 17k vs buying a transformer to get to 17k, will the tube operate at the same place...

It will tend to affect the bandwidth, especially the low frequency extension, of the power amp.
OTs are designed so that their primary inductance provides sufficient reactance such at full power output, the reactive impedance supports the intended impedance reflected from the secondary, down to the -3dB corner frequency that the designer is targeting.
Using a different load impedance on the secondary than that which the designer intended won't affect the primary inductance.
But it will change the frequency of that -3dB point, because the frequency at which the inductive reactance = reflected primary impedance  changes.
With the 8" speaker and tiny open back cab of a Champ, it may not be noticeable 
But if a big solid bass is required, especially a bass rig, mismatching the load may work against that.

[shooter]

but isn't that the point;
TGM want's to effect a change from "design center" to a more "musical" center, which has been "proven" with the OT swap.
my ? still stands, can TGM get that same "musical" center with the same speaker/cab..specs, EXCEPT 2X impedance.

[HotBluePlates]

"tube wants some particular load."

I think a light bulb lite
you're right, it's the designer that wants a tube to operate at a given point (tubes happy place:) ...

... will the tube operate at the same place ... speaker or OT? ...

The Transformer doesn't factor into the operating point (meaning idle bias) nearly at all.  The transformer determines where the tube goes, away from the idle point when signal is applied.

... my hangup is, using speaker load to get to 17k vs buying a transformer to get to 17k ...

I hear you; if one simply wants to see what happens in their amp with this-load or that-load, it's much easier to take a transformer you have on-hand and apply a different speaker-load to get there.  If you already have the different speakers, then it's also cheaper than buying a transformer.

How many math equations could have the answer 4?
2+2
5-1
28-24
√16
256 ÷ 64

Why does any transformer-maker have more than 1 4kΩ-primary transformer?  Why bother with "Vox style" and "Fender style" and "Marshall style" at all?  Why is there more than 1 kind of Champ OT?

There are a lot of different routes we could use to get to the answer "4" (way more than I showed).  There are also a lot of ways to wind a transformer to couple a 17kΩ primary to a 4Ω secondary.

• √(17kΩ/4Ω) = 65.19 : 1
• Could be 326-turns primary to 5-turns secondary
• Could be 1043-turns primary to 16-turns secondary
• Could be 15646-turns primary to 240-turns secondary

Each of the choices above will lead to differing amounts of primary inductance, leakage inductance, and interwinding capacitance.  And these are the things that shape the transformer's treble & bass response.  The transformer-maker might then choose among various schemes to sectionalize, layout & connect the windings to balance the tradeoffs.

There are as many design-tradeoffs in transformer design/manufacture as in the rest of a guitar amp.  The choices made for how to get to "17kΩ:4Ω" alter the resulting sound & frequency limits of the transformer.  There's a whole section of RDH4 (in the Library of Information) that gets into transformer considerations, design & construction that explains how complex it can be.

[shooter]

thanks HBP, I like simple, I already did "how many ways" math way back when, not going back 
My interest is more, can I quickly demonstrate to a guitarist "that tone" by adding dividing the speaker, then let that guide me to a "real" solution


The "where the tube goes" part I usually rely on the OT datasheet that has nice Log plots, the tube loadlines I use for "Dude, running like that, you'll need new tubes every other month"    

[pdf64]

"tube wants some particular load."

I think a light bulb lite
you're right, it's the designer that wants a tube to operate at a given point (tubes happy place:) ...

... will the tube operate at the same place ... speaker or OT? ...

The Transformer doesn't factor into the operating point (meaning idle bias) nearly at all.  The transformer determines where the tube goes, away from the idle point when signal is applied.
...

For an amplifier, isn't an operating point at idle something of a contradiction in terms, ie can an amplifier that's got nothing to do be thought of as operating

My way of viewing this topic is that the operating point is a signal (ie power output) dependant thing, ie as signal level rises from idle, the operating point will tend to shift. eg plate and screen grid voltages will reduce, bias (ie magnitude of Vg1-k) tend to increase.
Granted the change will tend to be more pronounced with AB amps that the class A Champ here.

[HotBluePlates]

... I like simple, I already did "how many ways" math way back when, not going back  ...

The "where the tube goes" part I usually rely on the OT datasheet that has nice Log plots, the tube loadlines I use for "Dude, running like that, you'll need new tubes every other month" ...

If you're averse to algebra, you're stuck with picking a load listing for a published condition on a data sheet.  Except most guitar amps don't conform to data sheet conditions...   

... My interest is more, can I quickly demonstrate to a guitarist "that tone" by adding dividing the speaker, then let that guide me to a "real" solution ...

How do you figure out the solution for the guitarist's situation?  Trial & error is lengthy, and maybe inconclusive.

... My way of viewing this topic is that the operating point is a signal (ie power output) dependant thing, ie as signal level rises from idle, the operating point will tend to shift. ...

Gotta nail down the simple before moving to "complicating factors."  Shooter seemed to think changing the OT primary impedance would alter the idle current of a cathode-biased tube (unless I misunderstood him).

Before diving into "signal-induced bias-shift" it would be good to hammer-home OT primary impedance does not cause the same changes as a preamp tube's plate load resistor.  IOW, the Champ 6V6 plate current would be very different if a 17kΩ resistor were between plate & B+, rather than a 17kΩ OT primary (with only 510Ω DCR).

[shooter]

changing the OT primary impedance would alter the idle current of a cathode-biased tube 

He did indeed, here's where I made the left turn;
looking at Tubewell's loadlines, Instead of looking right at the intersection of the 3 examples, I "followed" each example to the left side where each line "points" to a different plate current.  (I never did load-lines) I simply used the tube datasheet to find me a design happy place and built n tweaked to that. 


I can still do algebra, it's just not fun    and Fourier transforms are down right insane, best done by computer programmers, I'll just take the answer please 


on to dynamic analyses.....

[HotBluePlates]

changing the OT primary impedance would alter the idle current of a cathode-biased tube 

He did indeed, here's where I made the left turn;
looking at Tubewell's loadlines, Instead of looking right at the intersection of the 3 examples, I "followed" each example to the left side where each line "points" to a different plate current.  ...

Got it!  Now this, we can work with.

As I'm using the term, "operating point" is the same for all 3 loadlines:  340v plate, 250v screen (these curves are plate currents obtained when the screen is at 250v; high/lower screen volts results in higher/lower plate current), and -15.5v bias (however we got there).

Unless the 6V6 is driven by a transformer or a power amp able to drive current to the grid, the 0v grid curve is our boundary for max clean output power.

Now look along the light green loadline:
It runs into the 0v grid curve somewhere between 25 & 50 volts, maybe 38v.  So the 6V6 can only pull its plate as-low as 38v, or 340v - 38v = 302 volts peak.  Plate current rises from idle at 35mA to ~95mA, or 60mA peak.  CHECK: 302v / 0.06A = 5033Ω, so pretty close.

From here we can estimate the power output, 302v peak x 0.06A peak / 2 = ~9 watts.  EXCEPT! This is Class A, so the current-change for higher plate current should be equal to the current-change for lower plate current.  But 35mA-to-90mA is much more than 35mA-to-0mA, so we know the output tube will distort well before reaching 9w.

We can take a shortcut to power output by calculating the entire peak-to-peak plate current swing from 0mA up to 90mA: 90mA * 5kΩ = 450v peak to peak.  So 450v p-p x 0.09A p-p / 8 = ~5 watts and much more reasonable.

We would pull back in our memory-banks to know that a waveform squashed on one side has even harmonic distortion, and estimate this 5kΩ load has a lot of even harmonic.

Odd harmonic distortion is a sine wave squashed equally on both sides, which would happen when our drive signal runs into the 0v grid curve, as well as the 0mA axis.  The idle bias is -15.5v so the 0v grid curve is when the grid input signal is +15.5v peak.  The 0mA axis is not labeled, but is surely -40v or a bit more: 40-15.5v = -24.5v peak.  So distortion starts with a grid input of 0.7071 x 15.5v = ~11v RMS, and odd harmonic gets significant beyond 0.7071 x 24.5v = 17.3v RMS

You can also take a shortcut to the realization above by seeing the peak positive plate current happens when the grid receives a +15.5v peak signal (to reach the 0v grid curve).  So look to see what plate current does with a -15.5v peak, or the grid at -31v: the green load line is down to 4 or 5mA at -31v on the grid, but is not cut off.  This fact plus the dissimilar current-change indicates distortion, and a "non-optimum load."

Again, our lines & calculations are somewhat useless, because the actual amp has a screen voltage other than 250v.  We'd have to make our own custom plate curves using the existing graph and estimating plate current for the different grid voltages at a different, higher screen voltage.  RCA published an Application Notes on how to do that (back in the 30's I think).

[AmberB]

This is rather complex for my old grey matter...

[shooter]

It gets better and there are not many spring chickens here 

Odd harmonic distortion is a sine wave squashed equally on both sides,

so we've taken a perfectly good SE design and gave it a PP attitude, would be easier to have one of each

[HotBluePlates]

Odd harmonic distortion is a sine wave squashed equally on both sides,

so we've taken a perfectly good SE design and gave it a PP attitude, would be easier to have one of each 

Understand, though, it's adding odd harmonic to the existing even harmonic.

Any tube can exhibit even and/or odd harmonic distortion, even single-ended power or preamp tubes.  Push-pull just cancels the even harmonic distortion when well-balanced.

See the graph below for a range of load impedances for a single-ended 6L6... For the condition in the caption, a load around 4kΩ nearly eliminates 2nd harmonic, 4th harmonic is near its (relatively low) maximum, and 3rd harmonic is very prominent.  Interestingly, the 6L6 data sheet includes this exact condition but indicates a load of 2.5kΩ, lowering output power some but increasing the % of 2nd harmonic compared to 3rd and 4th.

[shooter]

Since I build mostly xSE i "add" evens, (flatten one lobe), in the pre amp, (usually 2nd gain stage) and "center" the PA.  I aim for ~ 1.5 - 2X signal needed to "push" the PA into distortion.