Speaker cone flutter from JBL K-120`s

[punkykatt]

Hey Guys,  A little over a year ago I built a Fender type combo amp about 25 watts, been using a  Cannibis Rex speaker with no problems.  Today I decided to try one of my JBL K-120 speakers and when I took the amp off standby the speaker cone started fluttering forcefully even with no signal.  I  hooked up my other  JBL K-120 and it did the same thing.  Reconnected the Cannisbis Rex, no problem.  Tried several other speakers and they worked fine.   I tried the K-120`s with another amp and they work fine.   Has anyone ever had a problem like this?       Thanks in advance for any help.  Punky

[FYL]

The Emi has a stiff suspension/spider translating into a high Fs of around 100 Hz - just like most guitar speakers, while the JBL uses a more compliant surround/spider and is spec'ed at 50Hz - typical of PA drivers who are designed to properly reproduce bass.

If your amp power supply pumps 60 Hz (poor filtering on the power tubes for instance) and the OT has large enough iron, it puts out a lot of sub 100 Hz garbage. The bandwith-limited Emi doesn't reproduce it, the JBL does.

[punkykatt]

Thanks FYL,  So if I want to use those JBL`s I need to increase the first filter cap in the PS?

[punkykatt]

I clipped in some additional filter caps to bump up the first node from 22uf to 82uf for a test.  Cone still flutters.  Do you think it may be the OPT? It came out of and old Hammond organ.

[HotBluePlates]

Does the amp use feedback around the output section? If so, try disconnecting it just for a test.

[FYL]

Does the amp use feedback around the output section? If so, try disconnecting it just for a test.

Yup. Flutter can also be caused by some forms of feedback-related motorboating.

[punkykatt]

Does the amp use feedback around the output section? If so, try disconnecting it just for a test.

Yes it does.  I will try that and get back later today.

[FYL]

Here's a 1927 article on motorboating. Nothing new under the sun.

[sluckey]

Another consideration... A perfectly balanced P/P output stage should perfectly cancel hum produced by power supply ripple. If the output tubes are not well matched, power supply hum will not be perfectly cancelled.  How well matched are your output tubes? Can you give us idle cathode current readings for each tube?

[punkykatt]

FYL,  thanks for that article on motorboating.  

Sluckey, Power tubes 7 mA miss match .034    .027.

I did disconnect the  NFB wire and the cone flutter stopped. But the amp now sounds noisy/scratchy, reverb squeals if turned up, and the amp has noticeable hum.  

[punkykatt]

I just checked that NFB wire I disconnected from the speaker jack for voltage, Im getting 26 VAC to ground, and 12 VDC to ground.  Is this normal? The amp has the LTPI.

[sluckey]

You're connecting the meter to the wire which is still connected to the circuit, right? If so, there should be some dc voltage present but no ac voltage. Sure you had 26vac and not 26mvac? Need to see your particular circuit to say more.

[punkykatt]

You're connecting the meter to the wire which is still connected to the circuit, right?

Yes,  I used two different MM`s to be sure.  On the 200AC setting they read 26.7 VAC, on the 750AC setting they read 26 VAC.

This amp has VVR (like tubnit uses)controlling the PI and cathode biased PA and appears to be functioning  as it should, it ranges from 42 VDC to 450 VDC.   I tried pulling  the preamp tubes one  by one  starting with valve one, the 26 VAC was present until i pulled the PI tube.  That would show no preamp problems correct?

[punkykatt]

I stopped the  cone flutter by increasing the NFB resistor.  Just love those decade boxes.  Not sure if its the proper fix  but the JBL`s like a 47k rather than the 4.7k.   Does that make sense?

[PRR]

It's maginally stable at a sub-sonic rate. So low that stiff speakers hardly move, but the fuller-range speakers will respond.

Zero NFB would be brutal on the bright JBLs.

Less NFB is not a wrong answer. It will significantly improve stability. You get a more raw tube sound, what Leo was moving away-from in 1960, but what goes around comes around and less-NFB toobes is a great alternative to the heavy-NFB sound of transistor amps.

Decrease the coupling caps to the power tube grids. For a test, try 1/5th the values.... fer example change 0.05uFd to 0.01uFd. That might get very-very stable. It might also give nasty lowest-notes, because the driver is now unable to pump full bass to the power tubes. If so, split the difference.

If perchance smaller coupling-caps makes the motoboating _worse_, there's not a good answer. Instead of tuning the coupling higher than the OT's bass-cutoff, you want to tune it lower.... but that probably means absurdly huge (and costly!) coupling caps.

[punkykatt]

Thanks for the reply PRR.  Im running .1 uF caps to the PA grids. I will sub in .022 uF caps and see what happens.

[punkykatt]

Well I thought I had .1 uF`s in there. Apparently I had changed them to .022 a while back.  I reduced the .022`s to .0047 on the PA grids and the fluttering  came back even with the increased 47k NFB resistor.  I believe I took out the .1`s because the amp was bass heavy to start with now that i think about it.  Oh well.  At least I learned something about the JBL`s.  Thanks All for trying to help.

[PRR]

> I built a Fender type combo amp

What is it *really*? Schematic, parts-values?

> the amp was bass heavy to start with

Reducing the grid caps is not the only way, sometimes not the best way, to take some bottom out. When NFB gets the caps' bass-cut interacting with the OT's bass-cut, sometimes it may be better to build the power stage "flat" and lose some bass elsewhere, in the preamp.

[punkykatt]

What is it *really*? Schematic, parts-values?


Dang, I had the schematic on my computer. Must have got deleated some how.  I will have to make up another. Might take me awhile.   But for now its  a single channel AB763 input, tone stack, reverb(2 tube), the gain stage triode after the reverb has the VibroChamp tremelo circuit wiggling the cathode,  LTPI into two 7591`s cathode biased.  Transformers from an old Hammond organ.  Also using VVR like Tubenit uses on the PI and PA.  

[PRR]

What a bunch of stuff.

Start with power supply chain. You certainly can't power all those stages from one power node, they would leak-back into each other and howl/shudder from birth. What I suspect is that you have just-barely enough power-chain filtering to suppress outright oscillation. A list of those R-C values and what stage(s) each supplies might be a place to start.

[punkykatt]

PRR, please excuse my sloppy hand drawn schematic of the power supply, PI and PA.  Please let me know if you need more info.  Thanks for your help.

[tubeswell]

Try a 100k NFB resistor (instead of the 47k)

[HotBluePlates]

Another consideration... A perfectly balanced P/P output stage should perfectly cancel hum produced by power supply ripple. If the output tubes are not well matched, power supply hum will not be perfectly cancelled.

But to actually see the cone move, it has to be pretty slow. Around 16-22Hz will result in visible pumping, but up around 60Hz, it's moving too fast. Hell, at 24 and 30Hz, it's moving too fast (which is why you don't see the screen flicker watching a movie or TV). So it's got to be sub-sonic oscillation.

Could it be the MOSFET in the VVR oscillating?

Make sure the components in that VVR are a tight, compact package. MOSFETs and transistors have a lot of gain. A LOT of gain... If you think tubes can oscillate easily, you're in for a surprise.

[punkykatt]

The 47k NFB resistor was large enough to stop the flutter.(see previous post)  But now when the reverb pot is turned past half way the amp volume decreases.  I reconnected the EMI speaker and the 4.7k NFB resistor the amp and reverb work normal.

I put the JBL back in and started process of elimination: Check all grounds and solder joints, Check resistors in PS, Clipped in additional filtering (40uF) on each filter cap in PS one at a time, (no change in flutter).  Disconnected VVR and temp rewire for non VVR operation, (no change in flutter).  Remove all preamp tubes except PI tube, (no change in flutter) subbed in a different OPT(no change in flutter) Pull PI tube flutter stops.  Got some negative vudoo in there somewhere???
 

[PRR]

Sorry I didn't see this for some days.

Everything looks reasonable.

> excuse my sloppy hand drawn schematic

No, it's beautiful.

The 7591 has almost twice the gain of the 6L6 family. That makes it more susceptable to sneakage; also the driver does not have to deliver real high levels.

So as one thought: change the B-C dropping resistor to something much higher. 17V drop is quite small. (Remember that in traditional amps this resistor also carries all the preamp current, you've put that on another path.) I think with 12AT7 and 7591, point C could be as low as 200V and still smack the output with authority. The B-C resistor could be 27K or 33K for ~~100V drop, 300V to driver. (A 1/2W part will do for tests; for long life it should be 1W-2W.)

Is that a 3H choke? 10H is often used here. In particular, 3H+22uFd resonates at 20Hz!! This is awful close to the 39Hz of the 0.022u+220K grid network. Also the 45Hz resonance of the big JBL. In theory, everything is push-pull so power line stuff cancels. In fact we have the 82K:100K unbalance, plus +/-20% tube tolerance.

Upping the C node cap even 4:1 only shifts this resonance down 2:1, so cap-change has poor leverage.

As a test, I would replace the 3H choke with 2K resistor (1W for test, 5W for life). If that makes it stable, there's your culprit. If it still makes full power and good tone, I'd run it that way. (Saves a pound!) A cathode-bias amp "shouldn't" have large current variation, thus little "choke bounce" to flavor the pluck transient. If it really needs the choke for tone, 10H with 40uFd (8Hz) might be a better tuning.

[punkykatt]

PRR, thanks for getting back. The choke is Doug`s bigger of his two Fender chokes. Not sure what the H is? I forget how
I came up with the 3H.  I will try those experiments you suggested.

[HotBluePlates]

The 125C1A is a 4H part.

4H and 22uF in a series-resosnant circuit resonates at 17Hz.

So either 3H or 4H lands you right at the bottom of the audio range, or in subsonic territory. That would leave you seeing the speaker pumping.

[GroundhogKen]

When this thread has played out, I vote it go into the archive.  Good stuff.


Ken  ( a big fan of G12H30 style speakers with 55 Hz resonance as well as JBL D120F )

[punkykatt]

HBP, thanks for the info on that choke.  That`s got to be the culprit.  Hopefully I will get a chance this weekend to do those experiments PRR suggested.

[Willabe]

4H and 22uF in a series-resosnant circuit resonates at 17Hz.

So where does 10H and 22uF end up?


            Brad     

[HotBluePlates]

If you follow the typical arrangement of power supply components and start immediately prior to the choke, the choke and the following filter cap are in series to ground.

So use the series resonance formula. C is in Farads, and L is in Henries. Look about mid-way down the page for f_resonant.

22uF and 10H are resonant at 10.73Hz.

Usually you don't have to worry about this stuff, but smallish decoupling resistors can allow strange interactions to happen. And it usually takes same/similar resonances in several places to cause problems. Like a perfect-storm situation.

[Willabe]

OK, thanks HBP.

I'll read the link and put in my favorites.


                  Brad       

[punkykatt]

Well I disconnected the choke and connected a 0 to 47k 5w power pot substitution box set at 2k. (no change in flutter unless I turn it up past 30k, dropping the voltage to the PI and screens way low). I Reconnect the choke and  removed the B-C node resistor, connected the  0-47k power pot sub box set at 33k. This reduced the flutter quite a bit, I set it a 47k and it reduced the flutter some more but still there. I tried clipping in additional Filtering on all the nodes (no change) PI still had some balls to push the 7591`s.  I don`t have a 10H choke on hand to try.  I have located a 9H @120mA for a reasonable price. Is that close enough?

[PRR]

> B-C node resistor... 47k and it reduced the flutter some more but still there.

Surprised. I'd expect flutter to just _stop_ at some point before 47K.

> located a 9H @120mA for a reasonable price.
 
You only need 30mA. It's powering screens and driver, not the big plates (and not even the preamp).

Something odd is happening. I can't figure what.

[punkykatt]

PRR, I forgot to mention that the slight flutter with the 47k in the B-C nodes takes about 30 seconds or so to get started.  Does that info help?  Should I still look for a 10H 30mA choke?