PI balance and ghosting - WITH CLIPS!

[fretwizard]

I've decided to open a new thread for my ghosting problem. instead of continuing on the thread where I was checking the voltages. I thought it was a bit confusing .

I've put together my Ceriatone 1959 in December:
http://forum.metroamp.com/viewtopic.php ... 1&start=45" onclick="window.open(this.href);return false;

and checked the voltages here:
http://forum.metroamp.com/viewtopic.php ... 3&start=15" onclick="window.open(this.href);return false;

Filtering is 50/50 F&T caps allround (formed according to Larry's Method). Wired like this: http://www.ceriatone.com/images/layoutP ... iatone.jpg" onclick="window.open(this.href);return false;

Larry grounding
3H Classic Tone choke, which was excluded by some, since the current rating is too high for saturation.
JJ EL34, biased at 67/68% plate dissipation, tubes are no more than 6mV apart in bias.
Also tried another matched quartet of E34Ls, still 6mV apart, still the same ghosting.

It seems to be the usual ghost note happening between strings D to B around frets 11-14, or at least on all notes in the same octave range. Neck pickup is more obvious.
I tried 2 different guitars, you can listed to clips here:
https://soundcloud.com/overtone85/sets/ ... host-notes" onclick="window.open(this.href);return false;

before installing a master volume I could say the ghosting is almost gone with Bright Volume at 3.
After I installed a LarMar PPIMV the ghosting is almost completely gone with master volume at 5.

I've measure the tail and cathode resistors at the PI and they all seem consistent in values, withing spec.
The plate resistor instead were 112k and about 91k.

I decided to try and balance the PI to see what happens.
First of all I replaced the plate resistors with a 100k and 82k. This made the amp a bit less gainy at volume 5 somehow. (I compared it to a clip I recorded once the amp was finished).

I've found Novosibir's post about balancing using resistors in parallel with the 82k and checking the signal voltages at the PI coupling caps, while feeding a 1kHz sinewave. (all controls at 10, presence 0)

At first I increased the signal volume until the PI output was 30vac on the 82k side, but there wasn't much difference between the 2 PI halves.

One day I was measuring the voltages out of a dummy load and I realized that a guitar signal was much higher, so I retried the test again.

Bright volume 10, tonestack 10, presence 0, LarMar MV 10. I plugged my Charvel with Duncan JB and a nice heavy strumming could deliver up to 70vac at the 82k side

I immediately hooked up a signal generator, 1khz, -19dB from my computer. This delivered about 68vac on the 82k side.
The 100k side measured about 57vac
I also remembered that I was testing the amp with a 12AT7. So I kept the signal volume as it was and swapped the PI with a 12AX7 and:
82k --> 57vac
100k --> 46vac

I connected the two coupling caps together to check the signal difference and it measures about 13vac.

I started paralleling big resistors on the 82k (values from 4.7M to 470k) to try and lower the voltage difference, without much luck.

When I tried on the 100k side instead I was able to lower the voltage difference down to 5.3vac, but I had to use a 34k resistor!
as I started to use even lower values the difference was going up again.

This way the ghosting is not completely gone but it seems reduced somehow. This resistor value is probably a bit too much since it's also reducing the PI volume quite a bit... and adding what sounds like 'compression? you should be able to A/B the tracks.
This might be all wrong but I thought I could be worth trying...

I've recorded two clips, similar to the ones above, same controls settings and neck pickup.

with no resistors added:
https://soundcloud.com/overtone85/charv ... comparison" onclick="window.open(this.href);return false;

With two 68k resistors paralleled on the 100k plate resistor, thus reducing the difference voltage to less that 50%.
https://soundcloud.com/overtone85/charv ... terbalance" onclick="window.open(this.href);return false;

Please let me know what you think. I've read somewhere there are other ways to balance the PI, on the tail resistor for example. Any suggestion on what I could try there?

I don't wanna increase the filtering...

[fretwizard]

I forgot to mention that I also tried to use 2.2k screen resistors, first on the 82k side of the power tubes and then on all tubes. The ghost wasn't really reduced and the tone changed...

Also my OT CT was twisted with the primaries. I untwisted it and moved away, didn't fix ghosting.

[fretwizard]

I checked Larry's post again and I realized it's impossible to balance with a signal voltage higher than 30/35vac, since the PI is overdriven and the negative feedback not working anymore.
I'll try the voltage divider method on the 82k next.
R

[fretwizard]

Alright... I was testing the PI output with a scope.
With ~30vac output at the cap, the two sides are pretty much symmetrical, only ~0.5vac difference.
With a guitar and all controls on 10 I could reach ~56vac, so I increased the volume on the signal generator.

At this volume, the 82k is between 9-11vac higher than the 100k in output, but the 100k has more clipping.
When I paralleled the 34k on the 100k resistor, as i the test clip i recorded, the 100k side has no clipping but also half of the volume and the 82k side of the power tubes starts red-plating.

Using the voltage divider method to lower the output on the 82k is perfect for matching the ac output but it doesn't cure the ghosting at all.
Anyway if you're interested, measure the ac voltage difference at the PI coupling caps (with a true RMS meter) and calculate the resistor values for the correct voltage drop. I did it here:
http://www.raltron.com/cust/tools/voltage_divider.asp" onclick="window.open(this.href);return false;
Replacing the 82k with a 72k-10k I could get the 2 outputs ~1vac apart.

At last I tried the MarshallHead suggestion, from this thread:
http://forum.metroamp.com/viewtopic.php?f=3&t=9192" onclick="window.open(this.href);return false;
Basically unsolder the PT CT and parallel the mains filter caps.
This did cure the ghosting, almost completely. You really have to pay attention when playing a neck pickup now.
The tone doesn't seem to change much either. I hope to post a clip soon.

Something interesting happened:
with 50uF on the mains, the 'matched' JJEL34 bias had a maximum mismatch of 6mVdc, the hottest tube being biased at 33-34 mV, the coldest at about 27-28.

with 200uF on the mains, the power tubes bias became perfectly symmetrical! 33mV on the inner tubes and about 29-30 on the outer tubes. Can anybody explain me why?

I got 50+50 F&T 500v caps and I biased at 62-63%, with a 485v idle plate voltage. I hope I will be fine. Otherwise I can still replace the caps with the original Ceriaton HolyGrail caps tha came with the kit, those are 600v! Do you think I should do that?
Any suggestion would be very welcome.

R

P.S. the first time I tried the amp with more filtering, one of the EL34s died and took both fuses (HT fuse exploded! stinky!). So now I've put the E34L quarted I had when I built the amp. They were 6mV apart on the bias but now they're much closer.
I didn't scope the PI output after this mod yet. I wonder if the combination of this mod and PI AC balance would remove the ghost completely...

[harddriver]

For some reasons some amps ghost more than others especially with low filtering like a 12000 series amp with 16uf on the screens. Here is someting you can try and it is something I used on my 12 series metro that really helped reduce the ghosting to where it isn't a concern to me now( almost inaudible). On the 12 series amps you need to add either 220K or 330K metal oxide balance resistors over top each of the screens filters caps + and - ends. I'm not quite sure how it works but I think it forces the screens filter caps to equally divide the voltage thus resulting in less ghosting.

On your 1959 I think I read you have 50uf filtering on the mains and screens. With that filtering ghosting should be minimal but try the resistor trick on your screens filter caps. If it works great if not pull them out. Just suggesting something simple that worked great for me but YMMV. I see they noramlly have 56K 3 watters onthem so maybe try upping the value.

DO a search on ghosting you might find more suggestions besides mine and maybe a better thread explaining things some more detail.

Good luck!

[fretwizard]

Hi harddriver

Actually snipping off my screen bleeder resistor is the first thing I tried, and I still haven't put them back in. Didn't help anyway but this thing of the 200uF on the mains helped a lot. The ghost almost disappeared and It didn't alter the tone much and the power tubes are almost perfectly matched in bias...
Gotta remember to record a clip ASAP.

regards

R

[fretwizard]

Interesting clues...

I found a guy on soundcloud who recorded test clips with many original marshall models.
You can hear an AFD ghosting right here, wait for the Sweet child o mine lick:
https://soundcloud.com/pleximaster/afd-100-no-epa" onclick="window.open(this.href);return false;

[Roe]

try this for balancing the PI:
decrease 1m on v3a to 390k ala park 1210 and/or increase NFB. 390k works well with 100k@4 ohms. with more feedback, you can increase the 390k almost to 1m