Weird Tremolo Problem, Maybe OT Impedance Mismatch?

[DougGuy]

I finally got done with the 5g9 build I started in the spring, brought it up on a variac with light bulb, everything checked out, voltages where they should be, no obvious newbie mistakes.  Keyword obvious..

The amp sounds really good, but I have this problem with the depth control of the tremolo circuit, as I turn up the depth, the speaker moves in and out with the bias wiggle surges.  The tremolo even sounds good just not intense in the depth.  I went over and over and over everything, and the only thing I found was that I might have used the wrong OT.  This was right after CT shut down and the OT I wanted was out of stock already, but I found one that looked like it would work, so I used it.  It has a 6k center tap. 

All the schematics, layouts, all of Rob's pages, they all show the pri and sec windings, color codes, etc, but NONE that I could see specify the input impedance of the OT!  I stumbled on this looking for the obvious, and found the list on Hammond's site for tube amp transformers listed clearly all of the specs for each transformer and also what application they were for.  Apparently the OT for the tweed 5g9 Tremolux uses a 4k or 4.2k input impedance, and Hammond lists their 1750J (1760J if you want multi taps) at 35w, 4k input impedance. 

I was not aware that the 5g9 was that much different than the 5e3, which uses an 8k input impedance on the OT primary, so I can't use that OT either, plus it's only 15w.

Would you suggest I get the Hammond OT with 4k center tap and then see if it works correctly, or is there a way to use the 6k center tap OT I already have?

There is an album of pics on google, there is a video in there of the weirdness this tremolo is doing:  https://photos.app.goo.gl/kkFBz4c5BThYmZVTA

[Latole]

IMO with 6K OT primary your amp should work.
You need to adjust the bias, you need a new bias.

Tremolo issue is often a wrong output tubes bias.

Read what a forum member, tubeswell ,do for bias;

https://music-electronics-forum.com/forum/amplification/guitar-amps/theory-design/6993-adjusting-bias-in-5g9

[HotBluePlates]

... 5g9 build ... as I turn up the depth, the speaker moves in and out with the bias wiggle surges.  ...

Trem on the 5G9 is injected at the bias-feed resistors running to the 6V6s.  You shouldn't get speaker-pumping unless your output section is really un-balanced.  Are your 6V6s somewhat matched?

... 5g9 build ...  The tremolo even sounds good just not intense in the depth.  ...

This amp is fixed-biased.  If you bias the 6V6s "hot" then the trem has a hard time shutting them off.  Bias the 6V6s cooler, and the trem will sound more intense.

... 5g9 build ... I might have used the wrong OT.  ...  It has a 6k center tap. 

Doesn't matter, your output transformer is almost certainly not a problem.

... Apparently the OT for the tweed 5g9 Tremolux uses a 4k or 4.2k input impedance, and Hammond lists their 1750J (1760J if you want multi taps) at 35w, 4k input impedance. ...

Hammond is WRONG.  They list the 1750J for the Tweed Tremolux but also for the blackface Tremolux.  In the blackface years, "Tremolux" was a 2x 6L6 head making 35-40w using a 4kΩ OT.  But in the tweed years, it was a "Deluxe with tremolo" using the Triad 108 OT.

Google "Triad 108 output transformer" and you'll find it's an 8kΩ part meant for 2x 6V6 and ~15w.  The 5G9 Tremolux schematic specifies the 108 OT, but your searches will associate it with the tweed Deluxe.  Mercury Magnetics' tweed Deluxe OT notes the original part # is 108.

Classictone's site would have lead you astray, as they didn't acknowledge the "Tremolux" name applied to anything other than the 40w blackface amps.
https://i.imgur.com/cre8PBI.jpg

[DougGuy]

... 5g9 build ... as I turn up the depth, the speaker moves in and out with the bias wiggle surges.  ...

Trem on the 5G9 is injected at the bias-feed resistors running to the 6V6s.  You shouldn't get speaker-pumping unless your output section is really un-balanced.  Are your 6V6s somewhat matched?

They were new JJ 6V6S both boxes marked 26 so they were matched.  I will trace down all of the connections to the octal sockets and make sure I haven't overlooked something I did wrong.

... 5g9 build ...  The tremolo even sounds good just not intense in the depth.  ...

This amp is fixed-biased.  If you bias the 6V6s "hot" then the trem has a hard time shutting them off.  Bias the 6V6s cooler, and the trem will sound more intense.

... 5g9 build ... I might have used the wrong OT.  ...  It has a 6k center tap. 

Doesn't matter, your output transformer is almost certainly not a problem.

... Apparently the OT for the tweed 5g9 Tremolux uses a 4k or 4.2k input impedance, and Hammond lists their 1750J (1760J if you want multi taps) at 35w, 4k input impedance. ...

Hammond is WRONG.  They list the 1750J for the Tweed Tremolux but also for the blackface Tremolux.  In the blackface years, "Tremolux" was a 2x 6L6 head making 35-40w using a 4kΩ OT.  But in the tweed years, it was a "Deluxe with tremolo" using the Triad 108 OT.

Google "Triad 108 output transformer" and you'll find it's an 8kΩ part meant for 2x 6V6 and ~15w.  The 5G9 Tremolux schematic specifies the 108 OT, but your searches will associate it with the tweed Deluxe.  Mercury Magnetics' tweed Deluxe OT notes the original part # is 108.

Classictone's site would have lead you astray, as they didn't acknowledge the "Tremolux" name applied to anything other than the 40w blackface amps.
https://i.imgur.com/cre8PBI.jpg

I have a 15w OT but I'd rather not use it if the OT isn't the problem.

[Latole]

OT is not the problem.
Check and adjust 6V6 bias

[DougGuy]

OT is not the problem.
Check and adjust 6V6 bias

I'm not ignoring you, that's my immediate next move, waiting on a small piece of eyelet board from Doug to mount the 1w 1ohm 1% resistors on.  another builder suggests using limiting diodes in parallel with these 1w resistors with cathodes going to ground.

[sluckey]

waiting on a small piece of eyelet board from Doug to mount the 1w 1ohm 1% resistors on.

I would just mount those resistors on the sockets, between pins 1 and 8. Remove the ground wire from pin 8 and move it to pin 1. Kinda like this...

[DougGuy]

waiting on a small piece of eyelet board from Doug to mount the 1w 1ohm 1% resistors on.

I would just mount those resistors on the sockets, between pins 1 and 8. Remove the ground wire from pin 8 and move it to pin 1. Kinda like this...

Doh I should have thought of that but hey learn something new every day.  Sluckey would you use limiting diodes along with the resistors?  Another guy who did a remarkably good 5g9 suggested the use of them in parallel with the resistors.

[HotBluePlates]

... would you use limiting diodes along with the resistors?  Another guy who did a remarkably good 5g9 suggested the use of them in parallel with the resistors.

Use them if you like.  They're used to keep the 1Ω resistors from burning (doesn't save anything else, though).

[PRR]

The diodes only act if there is 600mA for more than a moment.

That's not supposed to happen. If it does, I'd rather burn a 35 cent resistor than 35 buck tube.

Yes, factories do that. They don't usually warrantee the tubes. Also they may face nonsense "safety tests" where any smoke is a demerit.

[HotBluePlates]

The diodes only act if there is 600mA for more than a moment.

That's not supposed to happen. If it does, I'd rather burn a 35 cent resistor than 35 buck tube. ...

I made the same argument on another forum.  But the person advocating it showed a photo where his carbon film 1Ω resistors apparently burst into flame, burning some heater wiring along the way.

I wonder if a 500mA HT fuse would have been a better protective measure.

[Latole]

OT is not the problem.
Check and adjust 6V6 bias

I'm not ignoring you, that's my immediate next move, waiting on a small piece of eyelet board from Doug to mount the 1w 1ohm 1% resistors on.  another builder suggests using limiting diodes in parallel with these 1w resistors with cathodes going to ground.

Waiting your 1 ohms resistors you can try a colder bias with higher value in place of the 56 k

[DougGuy]

Thanks.  I have a 50kl pot there instead of a 56k resistor so I can adjust that value.  The board material arrives today but I like Sluckey's idea much better as it puts the bias test point right in your face more or less, very easy access with a meter, less proximity to high voltage.

The amp has been sitting the last couple of days I have learned a few things by reading the responses that I didn't know before so no harm to the amp and I am a little wiser than before.  Win win LOL.

I never learned how to bias an amp using plate dissipation I was never shown the importance of that method so this is all new.

[Latole]

You are going to do a very good job.

[DougGuy]

You are going to do a very good job.

Ha!  Easy enough for you to say 

Ok using the 50kl trim pot and r15k I can only get down to .038vdc at the 1r resistors with the trim pot turned ccw all the way.  414v at the plate.  Would going to a 22k in place of the 15k help that to go down more?

[Latole]

Schematic is easier to read for me.

The dropping resistor( arrow ) with higher value will send more negative voltage to tubes ( and less to ground ).
You will have colder bias.

[DougGuy]

Schematic is easier to read for me.

The dropping resistor( arrow ) with higher value will send more negative voltage to tubes ( and less to ground ).
You will have colder bias.

I changed out the 15k to a 22k, volts went from .038v to .034v, with the trim pot all the way ccw.  Putting the 15k and 22k together in parallel volts went to .043v, taking those out and putting a 10k resistor there volts went to .045v.  Correct me if I am wrong, but I need to get the reading at the 1r down to around .018v?  Given the above changes to the dropping resistor, what do I need to do now to accomplish this?

[sluckey]

BIGGER RESISTOR. That means go from 15k to 22k, to 27k, 33k, 39k, 47k, etc., etc.

[Latole]

Schematic is easier to read for me.

The dropping resistor( arrow ) with higher value will send more negative voltage to tubes ( and less to ground ).
You will have colder bias.

   Correct me if I am wrong, but I need to get the reading at the 1r down to around .018v? 

0.018 is colder bias, colder bias need resistor with more ohms (with higher value)

[DougGuy]

BIGGER RESISTOR. That means go from 15k to 22k, to 27k, 33k, 39k, 47k, etc., etc.

Worked my way up to a 56k and with the trim pot all the way ccw I can get r1 down to .019v.  The trem depth sounds correct but the bias does sound cold, backing the trim pot up some v4 reads .020v and v5 reads .024v at r1 respectively, and the amp sounds pretty good, still a little cold sounding to the ear but it's much better overall, could probably stand to go to .022v and call it good.  Swapping the 6v6 tubes balances the r1 to .021v and .022v at v5 and v4.  Plate voltage is 440 ugh..

The speaker is still pumping with the trem but not nearly as strongly as it was when the r1 was reading .045v.  At idle if I measure the r1 at v4, it's quiet, however if I measure r1 of v5, I can hear static "farting" in the speaker in time with the trem cycles while the probes are making contact with the checking points.

The amp sounds really good on .029v at r1 at v4, but only reading .022v at r1 on v5 so I guess the tubes are not as matched as the boxes claimed they were.  430vdc at the plates but love how the amp sounds. 

With the Altec 417 and a killer tricked out Les Paul with stainless frets and a Historic Makeovers finish, my own custom wound PAFs this tweed amp *nails* the tone Martin Barre gets on Aqualung.  Palm muting and lighter pick strikes gives the muddy Rio Grande tone of ZZ Top's first one.  All the nuances are there, this is 1970 to the tiniest detail. 

[HotBluePlates]

... down to .019v.  The trem depth sounds correct but the bias does sound cold ...

Therein lies the challenge of bias-vary tremolo:  cooler bias gives stronger trem, some people want the sound of warmer bias.  It's a balancing act.

[DougGuy]

Any idea what is causing the speaker to pump along with the depth control? 

[sluckey]

Try a 1N4007 diode across the outer lugs of the DEPTH pot.

[DougGuy]

Try a 1N4007 diode across the outer lugs of the DEPTH pot.

Thanks Sluckey, might have known you'd come up with something.  The diode cut about 85-90% of the pumping, still there but almost negligible at this point.  I have had a lot of trem amps but never seen one do this at the speaker.

I have some Golden Lion 6v6 tubes coming and the seller whom I got the JJ tubes from said they will take mine back and test them, see if they can get a better match.  Other than this I'd say the stock 5g9 Tremolux amp is pretty much finished.

[sluckey]

I have had a lot of trem amps but never seen one do this at the speaker.

I've seen several pump the speaker cone. If it really bothers you, just cut back on the intensity a bit. Or use a speaker with a stiffer cone.   

[pdf64]

… I wonder if a 500mA HT fuse would have been a better protective measure.

The best, and pretty much only, bright idea Marshall ever came up with was introducing the HT fuse to valve guitar amps
I’m at a loss why many other manufacturers, and DIYers, don’t use them 

I think a F250mA HT fuse would be fine on a 5G9, eg between the HT winding CT and its 0V common return.

I’ve started a program of adding bypass diodes to the 1ohm cathode current sensing resistors on my amps, after seeing the mess made in Jeff’s amp.
If there’s no HT fuse though, I suggest to stick with 1/2W 1ohm and no diode, but use fusible type (see Mouser), or at least MF with flame retardant coating.

On another note, it’s great to see old threads, where folks who are now seasoned old timers were asking noobie queries; I think the interchange on a good forum allows everyone to learn something, or at least cast a new light on things.

OK, I’ll extend Marshall’s bright idea list to include the unbypassed cathode on clipping stages of overdrive preamps 

[tubeswell]

Read what a forum member, tubeswell ,do for bias;

https://music-electronics-forum.com/forum/amplification/guitar-amps/theory-design/6993-adjusting-bias-in-5g9

Facepalm

[pdf64]

Read what a forum member, tubeswell ,do for bias;

https://music-electronics-forum.com/forum/amplification/guitar-amps/theory-design/6993-adjusting-bias-in-5g9

Facepalm

You weren’t sure, so asked, and learnt (and continued to do so apace). That a cause for celebration, absolutely nothing to be embarrassed about 
Even though it now seems weird that at some point in your life, you didn’t know that stuff 

[Willabe]

No one is born knowing things. No facepalm. 

I think your looking at it somewhat backwards? What I see reading that thread is; 

Look at the questions you were asking, you were thinking and processioning a LOT of ideas. And very quickly.  And that was 2008, look how much you've gained since then on amps. 

You never reported back on the ear test? So? What did you end up biasing that amp at and how does it sound? 

[shooter]

seems weird that at some point in your life, you didn’t know that stuff

knowing and retaining are opposing battles for the gray matter space! 


I stumbled across some old notes from the late 70's Navy days.  apparently I was pretty smart at one time  

[pdf64]

…All the nuances are there, this is 1970 to the tiniest detail.

You’ve got me thinking - I’ve got an old 5E3 kinda build I was looking to upgrade to a 5E9/a, perhaps a 5G9 might be a better option 

[DougGuy]

So the matched set of 6v6s tubes were only matched on the outside of the box    One pulled 29mv at v5 the one at v4 pulled 22mv not too matched to me, sent them back and the seller says they are matched on their equipment at 400v.  Harumph.  Yeah.  Right.  Okay..  They sent replacements that are much better at 29mv and 28mv respectively.

I left the vintage 12AY7 in V1, but swapped some JJ gold into v2 and v3, the amp got a LOT louder, and I noticed I didn't get the "ticking" that matched the trem speed when measuring current at the 1r on v5.  Interesting.  Swapped in an 8th gen Shugyang in v3, the PI, to see if it dropped the volume some, and when I turned the amp back on, I am greeted with this really loud WOMP WOMP WOMP even though the depth was turned all the way down.  Shut the amp off and went to bed.  Ugh. 

So swapping the v3 PI tube has a BIG effect on the trem.  Whodathunkit?  I have a bunch of 12ax7 will have to try them at the PI until I find one that behaves.  Might have to put the JJ back in there.

Edit to add question..  Could differences in DCR at the OT cause the power tubes to draw different amounts of current?  Before I finished this build just out of curiosity I ohmed out the OT, and I remember now that it didn't read the same from the center tap to the blue and brown wires.

[Latole]

from Weber books, a matched tubes between from 7to 10 ma are more musical, less steril than a perfect matched.

And I often see m.a. may change with time. Tubes  may came +/-  matched

[HotBluePlates]

Could differences in DCR at the OT cause the power tubes to draw different amounts of current? 

Look at a 6V6 data sheet (page 4) and see the curves are nearly-horizontal.  Changing/Lowering the plate voltage by 300v only reduces plate current by 20mA.  Hardly any change at all.

Instead, the screen (G2) and control grid (G1) voltages have a much greater impact on plate current.  A 50v change of screen voltage (page 6) changes plate current the same 20mA, while it only takes a 5v change at G1 for the same 20mA plate current change.

Different plate voltage is mostly meaningless.

... I ohmed out the OT, and I remember now that it didn't read the same from the center tap to the blue and brown wires.

The same is true of almost every transformer ever made.

Most wind a coil, with later turns over earlier turns.  The # of turns matter most to the transformer's function.

Copper has a fixed Ω-per-foot, and those outer turns are longer than the inner turns.  One-half of the winding then has higher DCR than the other half.

[DougGuy]

Can't figure out what's causing the pumping of the speaker, but Sluckey's suggestion of using a 1N4007 diode across the legs of the depth pot tamed it down tremendously.

Now for something I never would have expected, I have a bunch of 12ax7s, so I started swapping them in at v3, the PI.  Almost all of the Shugyang tubes generated some type of audible noise commensurate with the speed knob of the trem circuit.  Some bumped audibly with no input plugged into the amp, some didn't bump but made a ticking sound through the speaker when using a meter to read the current at 1r on v5, and one made a VERY loud WOMP WOMP WOMP sound when the amp was turned on.  I immediately turned it off.

JJ gold did none of the weirdness, none of the ticking when reading current at 1r, the amp was VERY loud, too loud really, almost harsh when driven all the way but I found a pair of new soviet production Mullard 12ax7/ecc83 and put them in.  Bliss at last!  They sounded really good, had none of the weird trem noises, and were not so gainy that I couldn't push the amp to it's max and get wonderful tone and sustain with a good bit of that woody tone.  Lots of open airy highs, feedback begins to edge in on long sustain notes, it's perfect to my ears.   

Only drawback is the trem is really not there, I'm not sure it can be made to work with the amp biased at 29mA which is about 85% dissipation according to Rob's calculator.  Truth is, the amp sounds TOO GOOD to mess with it any further!  Maybe another build is in the future I am a sucker for a COOL sounding tremolo but this one's in the books, not bad for a first build.  Thanks to all here who helped me out!!