70's italian amp restauration/modification

Discussion in 'Amps/Cabs Tech Corner: Amplifier, Cab & Speakers' started by HH1978, Jun 22, 2019.

  1. HH1978

    HH1978 Member

    Messages:
    85
    Joined:
    Jun 9, 2019
    Hello,

    One of my bandmate just brought me an old Manhattan 6024 bass that he used when he was young and has been sitting in the cellar for the last 30 years, and asked me if I could bring it back to working condition.

    At first inspection, it is obvious that one of the EL503 is shot (white getter). It also appears that the output transformer has an open primary coil (no resistance on one side). El503 are nearly impossible to find, so I think it would be logical to choose another power tube, since I'll have to replace the output transformer anyway. Reducing the power output (compared to the original el503 arrangement) is not a problem for him.

    I'm thinking to go with 5881, for the following reasons :

    - Easily available and not too expensive
    - B+ of the amp is around 365VDC
    - Heather current draw is lower than for the EL503, so I wouldn't stress the power transformer.
    - The bias circuit might work without modifications (though I have some questions about that circuit)

    Of course, I'll replace all the electrolytics and check the coupling caps as well.

    But before doing anything, I have a few questions :

    1) Does the 5881's seem a reasonable conversion? And will the preamp or PI need some modification to drive them properly?

    2) The original circuit power supply has a high amount of filtering : 200uF + 2X40uF + 100uF (and additional 16uF + 8uF on the preamp supply. Should I keep these values, or could I reduce the first cap value to 100uF?

    3) I don't understand the bias circuit arrangement. Am I correct to think the 10k pot is a bias balance, and the 1k pot adjusts the bias itself? But if then, why such a small value?

    4) What does 'CR' mean (from output transformer to negative feedback resistor)? By the way, the value of the feedback resistor is different on the actual amp (56k instead of 22k). I was thinking to take the feedback resistor from the 4ohms secondary, so I'll probably have to change that resistor, am I correct?

    5) What's the function of the 15ohms resistor and the 4,7 cap going from plate to screen on the power tubes? I don't remember having seen that before.

    Here's a schematic from a FBT bass organ, which is virtually the same amp :

    [​IMG]
    Many thanks for your inputs!
     
    middlechainringguy likes this.
  2. Tron Pesto

    Tron Pesto Member

    Messages:
    746
    Joined:
    Feb 18, 2016
    Location:
    Mahwah, NJ
    Taking a quick look at the data sheet for the EL503, I think there is a lot of things to consider here. First, you've got the wrong socket - the amp has MagNoval sockets - real odd ducks (at least in guitar amps). They would need to be replaced.

    The output transformer primary impedance wassn't going to line up that well with 5881s (or similar 6L6 variants) - about 2.4K for the EL503s and about 4K for 5881s. So, in a way, it's all well and good that the OT is shot - you would have wanted to replace it anyway for different tubes.

    And it seems as you saw, the EL503s can be driven with a low voltage and the PI in the schematic reflects that. You would have to beef up the PI circuit to provide a higher voltage. That should be easy enough and shouldn't draw that much differently on the PT. Look at a blackface Princeton that shows how the Cathodyne PI is set up - that's your target (there are others out there to compare).

    And if you're at it, you'll want to change the grid stoppers (3.3K) to a more appropriate 1.5K.

    Yeah, they probably have that to help with bass response seeing that it is an organ/bass amp, but that's overkill any way. 100uf is fine. I would bet even with a bass or organ, it wouldn't be any different.

    It's the other way around. The 10K pot sets the overall bias level and the 1K pot is a balance pot. A nice circuit to have. You'll need to see what the negative voltage range high to low to see if it is in a usable range for whatever eventual tubes you might use. It could require changing the value of the 10K resistor connected to ground.

    Yeah - that's the negative feedback loop. Maybe CR stands for "cathode return"? I think you have A LOT of other things to deal with before you make any change here. Once just about everything else is done, you can decide if you want to tweak this.

    Hmmm - yeah - odd. A partially AC coupled screen grid supply? (Scratches head.) Maybe trying to mimic a sort of faux-ultralinear transformer tap? Possibly some sort of fly-back protection (not exactly sure how that would provide protection-just spitballing)? (I'm sure others might have a better idea of the function.)

    I think the amp can work with these in or out - when you eventually get everything else sorted out, you can try lifting them to see the difference in performance/sound.
     
  3. Tron Pesto

    Tron Pesto Member

    Messages:
    746
    Joined:
    Feb 18, 2016
    Location:
    Mahwah, NJ
    BTW - love that you spelled restoration as restauration - LOVE me some Italian food!
     
    Sam Sherry and Baxtercat like this.
  4. middlechainringguy

    middlechainringguy Silver Supporting Member

    Messages:
    1,607
    Joined:
    Apr 9, 2017
    This is a great project. Thx for sharing.
     
  5. HH1978

    HH1978 Member

    Messages:
    85
    Joined:
    Jun 9, 2019
    Thanks!

    LOL sorry, I tried to hide I'm a french speaking guy, but I've been uncovered :D

    Yes, of course, I've already two octal sockets waiting next to the amp.

    Great! Exactly the information I needed! I knew the EL503 were much more sensitive than most power tubes, but I wasn't sure how to compensate the PI.

    Regarding the grid stoppers, would that make a lot of difference? I know Fender uses 1,5k, but Marshall uses 5,6k and I've seen values up to 47k in some circuits. On the other hand, I noticed the maximum grid resistance in fixed bias is 0,5M for the EL503 and 0,1M for the 5881. I guess I should check the total grid resistance in the circuit, which I haven't done yet, but even then, I'm not sure 2k woud make a lot of difference.

    Got it! thanks for clarifying!

    I'll try and see. I was sure it was unusual.

    I'll work the filter caps tonight, as well as the OT (one reason for targetting the 5881 that I forgot to mention is that I have a spare 4,2K OT). Once done I'll be able to power up the amp and see what voltages I can get.
    I'll report once done :)

    Many thanks for your advices!
     
  6. HotBluePlates

    HotBluePlates Member

    Messages:
    6,228
    Joined:
    Mar 27, 2017
    I looked at an EL503 data sheet then thought: YEEOOWWWW!!!!!! The whole output stage will need to be re-designed, as will the bias supply.

    The EL503 has a Gm of 23,000 micromhos, or about 4x that of the 6L6/5881 (and more than double the Gm of an EL34). This means the EL503 only needs 1/4 the drive signal to make its output power compared to the 6L6, or half the drive signal of the EL34.

    The high Gm also means it requires less bias voltage to get it under control. Notice the data sheet shows a condition on page 2 with a low B+ voltage (under 300v), and a shared bias resistor of 56Ω (where 200-250Ω would be typical for a 6L6). We have fixed bias in your amp, but it's showing only -20.5v at the grid which may not be enough.

    Small bias & small drive means you'll need to be thinking about phase inverter design and the available B+ voltage. There's a good chance of having to switch to something like a paraphase inverter to get a big enough output to drive the 6L6s, but a lot depends on your design goals/compromises.

    Switching tube types means you'll need to re-design the entire output section/phase-inverter. I don't know anything about the OT to know if it might be re-usable.

    Feedback from plate to screen. You might even look at it as feedback across the half-primary connected to that tube.

    I'm not going to go too far into trying to guess the reason... Tubes with Gm that high have their screen grids perilously close to G1, and can easily oscillate. It might be phase-correction for the OT and the feedback loop, or an effort to kill parasitic oscillation, or both (or something else).

    Ordinarily, if I see a plate-to-screen connection I think "triode mode" and ultralinear is just a halfway point between pentode mode & triode mode operation. "Triode Mode" means the screen voltage falls/rises exactly the same amount & same time as the plate, is a form of local negative feedback, increases drive requirements, etc. But the screen is tacked right to a filter cap, so it won't fall much as the plate current/voltage varies.

    The EL503 is a low-voltage, high-current tube. The high values of filter capacitors are for sourcing those big current swings.

    If you change them, you now add power supply design to you list of chores for this amp. Again, depending on your goals, you might find yourself painted into a corner.

    You can just bolt-in whatever, and it will work "good enough" as long as you don't go crazy with negative feedback. It's your call about what constitutes "good enough" for the project.
     
  7. pdf64

    pdf64 Member

    Messages:
    5,906
    Joined:
    Aug 9, 2008
    Location:
    Staffordshire, UK.
    Maybe consider a quad of el84?
     
  8. HotBluePlates

    HotBluePlates Member

    Messages:
    6,228
    Joined:
    Mar 27, 2017
    I haven't read all of it, but This Guy decided to use EL86/6CW5 in place of the EL503 tubes (I don't know John, but the projects he's posted on his site show he's extremely capable in designing electronics & transformers). This is a smaller tube, like an EL84 replacing an EL34, but they are cheap NOS, have the Gm of an EL84/EL34, and will likely get the job done with the supply voltage available.

    The phase inverter will still need investigation after the output stage is determined.
     
    Last edited: Jun 22, 2019
    Rob s likes this.
  9. HH1978

    HH1978 Member

    Messages:
    85
    Joined:
    Jun 9, 2019
    I've seen this page before and this was my first idea, as he went through almost every other power tubes before settling on these ones. But I think he used EL36 and not EL86. These tubes have indeed a Gm of 14ma/v, so even more than the EL34/EL84, and the heathers draw about the same current as the EL503. The top caps should fit in the amp, so no problem on that side. My concerns though are the 250V max plate voltage, and the impedance (the datasheet doesn't show the operating conditions in class AB).
    They are about as cheap as the EL86 though.

    As for the EL86, it would also require to lower the voltage. Except the fact that they are cheaper, I'm not sure there would be a benefit vs using EL84.
     
    HotBluePlates likes this.
  10. HH1978

    HH1978 Member

    Messages:
    85
    Joined:
    Jun 9, 2019
    Well, partially only, as it's not my amp. So I spoke with my bandmate on the phone to be sure what he's expecting me to do.
    It's clear that he wants it to be working again mainly for sentimental reasons. It was his first amp and he used it throughout the 80's. Currently, he uses a hotrod Deluxe with our band and is satisfied with it.
    So he'd like to get the amp working, but without spending an awful lot on the project, and we both agreed that, given the low value of the amp and its actual condition, we should make some compromises.

    Changing the OT is a must. I have a spare 4,2K primary, 4/8ohms secondary that I bought for a Vibrolux reverb build, but never used as in the meantime I traded in a 1966 Vibrolux reverb. Hence my idea to use that. As it is arguably the most expensive part, it seems reasonnable to use it as the startpoint and design the rest of the amp around it. Tweaking the PI might need some thinking, but the parts are likely cheap. As for filter caps, I thought about using 100uF instead of 200uF because it's easier to find in the cap can form. CE makes the 200uF+40uF but it is really expensive. I could use single electrolytics, but I don't see where I could place them in the amp.

    By the way, my bandmate told me he had tried the amp before bringing it to my home. It powered up, and made sound, although not a very pleasant one. Sounds likely, as it worked on one tube only. Since he already powered it up, I guessed I could do it too without causing more harm, to see what voltages I could mesure. They are all very close to the schematics (except of course for the plate that is connected to the bad side of the OT), which sounds like a good news. One thing I'll have to take care of, though, is the heather voltage, a bit high (3,45Vac on each pin, so 6,90Vac total, well above the 5% limit). The mains selector is on 240V, and my wall supply is 233V, so I'll have to use resistors I guess.

    The measured bias voltage ranges from -9V to -29V. Not sure if it will be enough for the 5881's?

    I'm not familiar with the paraphase inverter. If I understand the concept correctly, the second half of the triode would be used as a gain stage for one half of the signal only, and attenuated with a voltage divider before being sent to one power tube, while the other power tube gets the other half directly from the preceding gain stage, is that correct?
     
    HotBluePlates likes this.
  11. HotBluePlates

    HotBluePlates Member

    Messages:
    6,228
    Joined:
    Mar 27, 2017
    Thanks for the correction! I must have seen what I wanted to see... :p

    Except he notes towards the bottom of his page that the tubes can handle quite a bit more than rated, and he runs them at 425v plate, 200v screen.

    No matter what, I'm thinking you have a design project on your hands, and mainly have to think about how much of the power supply & output section you want to replace to get it running.

    Required bias depends on screen voltage.

    The "triode amplification factor" or "Mu G1-G2" of a 5881/6L6 is 8. So a bias of Screen Volts / 8 will come close to cutting off the tube. If your screens are at 300v then 300v / 8 = ~-38v would be desirable as a max-negative figure. I'm think a bias circuit that varies from about -15v to -40v would be suitable to accommodate a variety of tubes.

    Working from the above, if -38v is somewhat near cutoff then Class A might be idling somewhere around half that, or -19v. Class AB might be a little cooler, like -22v down to actual cutoff near -40v.

    All this is obviously ballpark guessing, but gives you a sense of where you want to go.

    The first triode of a paraphase inverter amplifies the preamp signal enough to fully drive one side of the output stage. The second triode is only used to invert the signal polarity for the other side of the output stage. A voltage divider that reduces signal as much as the 2nd stage amplifies is used between these triodes.

    The advantage of a paraphase is you are developing only one output signal for the amount of B+ available (better than split-load/concertina that has to develop 2 output signals and still leave voltage across the tube), and you are not wasting voltage across a large tail resistance (as with a long-tail inverter).

    You will work to design the output stage first; once you know the bias voltage required for the output tubes using the given B+, you will know the peak output voltage the phase inverter must deliver to each output tube. The peak drive voltage required from the phase inverter and the available B+ will guide you as to what phase inverter circuit is feasible.

    A split-load inverter can generally only deliver a peak output voltage that's 1/4 (or maybe 1/3) of the total B+. That's because the plate output negative peak and cathode output positive peak happen in such a way as to leave minimum voltage across the tube. And a triode might need ~1/3 of the B+ (70-80v) left across it to amplify. So this means split-load inverters work well when the available B+ is high or the required drive voltage is small (or both).
     
    Last edited: Jun 23, 2019
    pdf64 likes this.
  12. Kyle B

    Kyle B Supporting Member

    Messages:
    4,882
    Joined:
    Sep 10, 2012
    Location:
    Western NY
    HotBluePlates likes this.
  13. HH1978

    HH1978 Member

    Messages:
    85
    Joined:
    Jun 9, 2019
    Thanks!
    But actually, I already replaced the sockets with octal ones, as they are likely what I will use, either for 5881 or EL36 (with an additional top cap in that case).

    I'm tempted to order a pair of EL36, since they are really cheap, as well as top caps for the plates. The idea of trying an unusual tube is very appealing.

    In the case I'd go for EL36, I'd have to reduce the screen voltage to around 200V (currently the voltage on pin 4 is 355V). Can this be done with a dropping resistor from B node, or is there a better way? As I understand, John Chambers used a separate power supply for the screens, but I'd prefer to avoid that if I can.

    I still have to take a closer look at the bias circuit. I could not make any progress today as I was busy with the family.

    @HotBluePlates, thanks you so much for the detailed explanations! That really helps me to define the guidelines for the project :)
     
  14. HH1978

    HH1978 Member

    Messages:
    85
    Joined:
    Jun 9, 2019
    I took a closer look at the bias circuit. It uses a separate winding on the mains transformer, at 25Vac, giving roughly -30Vdc after the rectifier diode. Trying with 5881, I couldn't bias them colder than 85% of max plate dissipation, so I shut down the amp and didn't go further.

    Would it be possible to tap the bias circuit voltage from the HT winding through a dropping resistor, like on a JTM45 for instance?
     
  15. HotBluePlates

    HotBluePlates Member

    Messages:
    6,228
    Joined:
    Mar 27, 2017
    You don't want to drop a bunch of volts with a resistor for output tube screens. That's because alternating/signal current will be more than a few milliamps, so the screen voltage will sag mightily with signal. Falling screen voltage means falling plate current, so you will constrict power and have a heavily compressed sound.

    The above is why John used a separate screen supply. If you want to ease screen voltage requirements while also having fairly easy drive, you might look into just using an EL34. An EL34 data sheet shows that idling at 70% with 365v on the plate and a screen voltage of 325v will require about -25v bias. A 25v peak drive signal should be reasonable with the power supply voltage available.

    FWIW, I was looking at the original schematic, and the voltages noted around the phase inverter cannot possibly be right. There is "160v" at the cathode of the split-load inverter, and 240v at the plate... But the same ~4.8mA has to flow through the plate load that flows through the cathode load, so the B+ for Node C would have to be 240v + 160v = 400v, or higher than the voltage feeding the output tube plates & screens!
     
  16. Tron Pesto

    Tron Pesto Member

    Messages:
    746
    Joined:
    Feb 18, 2016
    Location:
    Mahwah, NJ
    What you probably could do is build a voltage doubler circuit out of the bias tap. Google that and you'll find a bunch of examples. The bias circuit draws little to no current, so you can build a voltage doubler or even a tripler.

    Here's a link that may help: https://www.electronics-tutorials.ws/blog/voltage-multiplier-circuit.html
     
    pdf64 and HotBluePlates like this.
  17. HH1978

    HH1978 Member

    Messages:
    85
    Joined:
    Jun 9, 2019
    Thanks! I'll try to build this tonight and see what voltage range I end up with. A doubler should be enough.

    Taking careful measurements, it looks like the EL36 will be too high to fit in the cab... bummer...

    I just noticed TAD has some EL503 in stock, but at 300€ the pair, that's more than the value of the whole amp.
     
    HotBluePlates and Tron Pesto like this.
  18. Tron Pesto

    Tron Pesto Member

    Messages:
    746
    Joined:
    Feb 18, 2016
    Location:
    Mahwah, NJ
    Of course, keep in mind you will be building a NEGATIVE voltage supply - be sure you have your diodes and caps oriented correctly to provide a negative voltage. (That web site has examples that provide a positive voltage.)
     
  19. HH1978

    HH1978 Member

    Messages:
    85
    Joined:
    Jun 9, 2019
    Thanks for pointing that out! Reversing each of them should work, is that correct?
     
    Tron Pesto likes this.
  20. HH1978

    HH1978 Member

    Messages:
    85
    Joined:
    Jun 9, 2019
    The voltage doubler works perfectly :), I have now a range from -30Vdc to -60Vdc to adjust the bias. -33Vdc gives me a plate current of 42mA at 365V, so 66% of max plate dissipation for the 5881's.

    Now that the power tubes are running safely, I tried the amp to see how it sounds as is... As expected, the signal is weak, so the next move will be to redesign the PI, I guess.

    By the way, while I was at it, I tried several different tubes in the PI slot. With the ECC82 and ECC81, the signal was weak, but other than that, sounded "normal" to me. When swapping in a 12ax7, it was a bit stronger, but with a pronounced tremolo effect. Any idea of what could cause the oscillation? I doubt it is an effect of the tube itself, more likely something that is present in the circuit and gets amplified more with the 12ax7 than with the lower µ tubes, or am I wrong?
     
    HotBluePlates and Tron Pesto like this.

Share This Page

  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice