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
    I tried to build a paraphase inverter, starting with 5c3 and 5d3 schematics, but something went wrong, so I probably misunderstood how it should be done.
    Currently, it passes the signal, even a strong one, but also makes a strong "ground issue" noise.

    Voltages read : 170v on first triode plate and 2,5V on the cathode. That part seems about right. But on the second triode, I get full C node power on the plate, and 0V on the cathode. The voltage from C node has no other path to reach the plate than the 100k plate load resistor, and tracking the resistance, I have a correct 1,5k reading from cathode to ground, so I'm scratching head.

    Testing continuity also produce strange results : I have continuity to ground on each side of the 220k power tubes grid leaks, as well as through the 4,7k resistor, but only with the amp powered up. I don't know how to explain this. As it stays a while after powering the amp down, then disappears, I suppose it has to do with the tubes being hot and conducting.

    I'm not sure either about the first triode 1M grid leak. I guessed the pot on the 5d3 schematic acts effectively as a grid leak, but as I have a preamp gain stage after the pots, I wasn't sure what I had to do.

    I'm not sure either were to connect the negative feedback resistors, so for now the circuit doesn't have a NFB loop.

    Here's the schematic (I forgot to anote the pot, it's the bias supply balance) :[​IMG]
     
  2. Tron Pesto

    Tron Pesto Member

    Messages:
    746
    Joined:
    Feb 18, 2016
    Location:
    Mahwah, NJ
    I'm trying to follow what you did compared to the paraphase inverters I'm familiar with and it's not lining up. Now it could be a valid circuit - I do not claim to know all of the possible paraphase possibilities, but one thing jumps right out to me....

    UGGG - Edited for everyone's mental health - I've way overcomplicated things and probably made it worse

    Would love others to review and chime in on this.

    You can see other potential paraphase circuits here. Or why not go for an updated cathodyne, similar, but improved from the original?
     
    Last edited: Jun 30, 2019
  3. HH1978

    HH1978 Member

    Messages:
    85
    Joined:
    Jun 9, 2019
    I think you're right, as I said I dont feel like I'm understanding the paraphase correctly, as it is a totally new type of phase inverter to me. I tried to mix the circuits I saw in old deluxe schematics with the existing circuits in the Manhattan, but I was under the impression that I was wrong somewhere.

    Do you have a link to a paraphase schematics for a class AB fixed bias amp?
     
  4. HH1978

    HH1978 Member

    Messages:
    85
    Joined:
    Jun 9, 2019
    I've seen the Merlin schematics too, but am I missing something or the coupling caps do not seem to invert the signal?
     
  5. Tron Pesto

    Tron Pesto Member

    Messages:
    746
    Joined:
    Feb 18, 2016
    Location:
    Mahwah, NJ
    Yikes, you know, I may have lead you down the garden path... taking a Icloser look through the amp, I'm beginning to wonder if "you can get there from here". In other words, the problem is greater than which PI circuit will work, it is likely the gain structure through the entire amp is not going to deliver a signal strong enough to push EL34s.

    It's seems that the amp is designed to minimize distortion due to high-gain preamp signals - and that may make sense for Bass and/or Organ in their mind back then when designing it. To wit, that is probably why they used EL503s - they designed an amp that needed low drive output tubes.

    I'll take a closer look at the entire amp (shame on me for not digging in previously). Hopefully others could shed some insight too.
     
  6. HH1978

    HH1978 Member

    Messages:
    85
    Joined:
    Jun 9, 2019
    Don't feel sorry for that. My main motivation for taking this project is to learn, and learning implies trial and errors. By the way, my summer vacation (I'm a teacher) starts today, so I'll have some time to spend with the project :)

    I'm not sure it was actually the reason, since they made a guitar version of the amp, and that version used EL503 too. Seems like they bought in bulk, and made some compromises using the same parts in all their amps. I have two other amps that were probably made at the same factory, but a bit earlier. One uses a pair of EL84's and the other uses a pair of ECL82, with the triodes working as PI and the pentodes as power tubes. Both are combo, and are very underpowered given their overall size. The cabinets are made of MDF. Made on the cheap would be how I describe them, though they sound really good (especially, the bias tremolo is to die for). They usually came with Mullard or Philips Herleen tubes, and mustard caps too.

    Here's the preamp schematic for the guitar version of the amp :

    [​IMG]

    I don't have the schematic for the power section, as the guy that posted this on his webpage rebuilt his amp using new mains and output transformers, and using a tube rectifier, but I would bet it's the same as the bass organ version. I know for sure it used EL503's, as it is the reason why the guy had to rebuild it.

    I discarded EL34's from the start, as their heather's current draw is about 1A more than EL503's, and I'm not sure the power transformer would be able to handle that. But it seems clear that the preamp will have to be redesigned too. The OT I used is rated for 50W and has a 4,2K primary, 4ohms and 8ohms secondary winding, so good for 5881/6L6. I think 6V6GT could work too, using 1 step up speaker load (i.e. 16ohms speaker from 8ohms secondary), but I'd like this to be confirmed.

    Looking at the amp, I'm thinking it would certainly benefit from being rebuilt entirely. The wiring is on the ugly side (i.e. : heathers wires run everywhere through the amp and are not twisted together.

    Discussing with the owner, it seems clear that I have total freedom for the project, as long as I don't go through the roof with parts price. He doesn't even remember clearly how the amp sounded like back in the days, so staying close to the original is not even a goal.

    So, thinking about the actual configuration, how he would actually use the amp, what parts I have lying around...

    - I have two channels in the preamp, 2+1/2 ECC83, but there's physical room for at least two more tubes in that section if needed. My friend doesn't need two channels, so I'm free to reconfigure the preamp (i.e. : implementing a tremolo). A pair of EL503's need 2,2A for the heathers, a pair of 6L6 needs only 1,8A, so adding one ECC83 is possible, two would be an unsafe bet with the current mains transformer. So I could use 3 ECC83 + 1/2. More would be possible if using 6V6 power tubes.

    - Power output doesn't really matter. It's unlikely that he would use this amp for gigging, and even then, the amp would be mic'ed through the PA.

    Here are the options I think of :

    1) With the current power transformer :

    - Turning the amp into a 5F11 tweed Vibrolux. Unless I'm missing something, I could use the second channel's control for the tremolo, and the B+ would be sufficient.

    - Turning the amp into a Magnatone 213 (except for the solid state retification). Being an old Magnatone afficionado, I have some spare varistors, though I admit I would be a little reluctant to use them in someone else's amp. That said, I read the varistors can be emulated using zener diodes, and that could be something interesting to try. My own 213 runs with a 360V B+, and uses 3 preamp tubes (one 12AU7 and two 12ax7). The PI is a cathodyne configuration and uses a 12ax7.

    2) Changing the power transformer

    I guess pretty much anything could be done going that way, but I would be interested to try one of the Brownface harmonic tremolo designs.
    Real estate is not a problem. The chassis is as large as a JTM45's, and that's only for the transformers, power amp and PI, the preamp being on a separate board.

    Let me know what you think...
     
  7. pdf64

    pdf64 Member

    Messages:
    5,905
    Joined:
    Aug 9, 2008
    Location:
    Staffordshire, UK.
    Yikes, this thread has become rather convoluted, due to so many options and rabbit holes.

    I think you may be getting a bit muddled there; I can see where you're coming from but I'm a bit stuck trying to explain exactly what you've got wrong. Whatever, unless we're intending to change to class of operation, eg class A1 to class AB1, I don't think there's particular need to change the idle point.

    I don't think it good practice to have component leads / bodies so close to other eyelets / turrets; things may get moved inadvertently or insulation fail, and things may short. Just use the regular Fender / Marshall layouts as a guide / template.

    The bias for the power tubes is also being applied to V4b grid, which will put it deep into cut off; hence its plate will stay up at its HT supply node Vdc.

    See the last sentence from Merlin's page http://www.valvewizard.co.uk/paraphase.html
    if the power valves are fixed biased then the paraphase will needs its own coupling cap and grid leak from the junction of Rf and R1

    See the later Vox AC50 for such a paraphase https://el34world.com/charts/Schematics/files/Vox/Vox_ac50_1.pdf

    However, I think that a regular cathodyne would be ok, see the 5E7 as an example of a cathodyne driving fixed bias p-p 6L6 at a lowish HT https://el34world.com/charts/Schematics/files/Fender/Fender_bandmaster_5e7_schem.pdf

    Horrible things are going on in that soundclip, maybe motorboating causing the trem effect, and the distortion may be an operating point in cut off, or could be ultrasonic oscillation forcing a stage into cut off. Bear in mind that with either a 12AT7 or especially a 12AX7, the common cathode triode stage will have more gain that it would stock with a 12AU7; hence the likelihood of oscillation increases, either at sub audio via the HT (motorboating) due to insufficient decoupling, or ultrasonic due to parasitic capacitive coupling via lead dress, layout etc.

    Sorry if I overlooked it, but not sure what was done with the power amp's global negative feedback loop in the above configuration?

    I suggest that you start with by removing all tubes other than the power tubes, setting and reporting the power tube operating point, ie plate screen grid control grid Vdc and plate or cathode current.
    Then add a phase splitter tube and whatever circuit type you want for it (any should be fine), connect the input directly to that, and get the power amp operating nicely.
    Then add in a preamp stage or 2 to make sure that everything behaves itself / it can achieve a decent power output.

    Then consider what to do with all that rather complex original preamp.
     
  8. HH1978

    HH1978 Member

    Messages:
    85
    Joined:
    Jun 9, 2019
    Thanks for chiming in :)

    I took the voltages on the power tubes :

    Plates : 365Vdc
    screens : 366Vdc
    grids : -28,1Vdc on one tube, -29,7Vdc on the other
    heathers : 3,40Vac on each pin
    Plate current : 42ma on first tube and 38ma on the second one, so one tube is running at 66% of max plate dissipation, and the other at 60%. That's the closest I could get with the balance circuit.

    I noticed the sentence on Merlin's website, but that was after trying to build a paraphase. I suppose my wiring was wrong enough to explain bad sound. One thing I still can't figure out : do the coupling caps invert the signal? I know the triodes do when the signal leaves at the plate.

    I didn't know where to inject it, so it was left out of the circuit.

    I agree that I would have needed to make another layout for the long tail if I wanted to use it permanently, but it didn't work well anyway with the available voltage. I'll probably go back to a cathodyne.

    Thanks, that seems like a good starting point. I was thinking about putting the 470k grid leak on the amp's schematic (the actual value was only 100k), I see the 5E7 uses 1M. I was also thinking to increase the plate and cathode resistors to 56k or even 68k.
     
  9. pdf64

    pdf64 Member

    Messages:
    5,905
    Joined:
    Aug 9, 2008
    Location:
    Staffordshire, UK.
    No coupling caps don't change signal polarity.
    Outside of its passband, a single pole RC filter will shift the phase of the signal, to a max of 90 degrees, so that may be what you're thinking of, but that's not something we can use to change a single ended full range audio signal to a balanced one (ie what a phase splitter does).
     
  10. HH1978

    HH1978 Member

    Messages:
    85
    Joined:
    Jun 9, 2019
    Thanks, I thought so but wanted to confirm.

    I rebuilt the cathodyne, except I used 68k plate and cathode resistors, and a proper 470k grid leak. It sounds better than what I remember. Still a hint of fizziness as the notes decay though. I see Merlin suggests using a 100k to 1meg grid stopper resistor on the cathodyne triode to avoid ugly distortion. Should I try that? And if so, what do I do with the grid leak value? The example he provides uses a 470K grid stopper and a 47k grid leak in series with an arc protection diode.
     
  11. HH1978

    HH1978 Member

    Messages:
    85
    Joined:
    Jun 9, 2019
    The things seem to come into shape. As I was a bit desperate to make this amp sound right, I went through my stock of tubes to see if I had an option I didn't try before. I found some 6973's, but there's no way I would want to use that, as I will need them someday for my Magnatone 280. 7868's too, but I don't have the spare sockets... And then I found another output transformer on the shell. Not sure when or why I bought this, but it is a 6,4k primary, 4ohms and 8ohms secondary, rated for 40w. The one I used was 4,2k primary, rated for 50w.

    I thought that the impedance would allow me to try KT66's on the Manhattan. Typical impedance is 8K, but I know the early Marshall used a 6,6K winding, so why not? Datasheet for 5881 shows it can be used in class AB with 6,6k ptp, so both tubes could be used with this OT.

    The difference when I tried the KT66's was spectacular (compared to 5881 with the other OT). No more fizziness, much warmer tone and better definition. I also tried all the suitable tubes I had for the PI slot : Ecc82, 12AY7, 12AT7, 5751 and Ecc83. They all worked fine, no more motorboating or tremolo. The 12ax7 and 5751 have too much gain, though. I think I like the 12AY7 the best, but the ECC82 and 12at7 work very well too, different flavors...

    The KT66's draw about 1,3A for the heathers, the original EL503 about 1,2A. I don't know the power rating of the mains transformer, but I suppose an additional 200ma should be ok. The transformer didn't seem to get hot.

    At this point, I was thinking that the KT66's were much better suited to the amp than the 5881 I had tried before. Anyway, I decided to try them again with the new OT. Albeit very different than with the KT's, the sound was good. More defined and percussive, a little less warmth. And again, all PI tubes worked without issues (except the excessive gain for some). I can't really decide which power tube I prefer, so I'll leave the final decision to the owner, and anyway, as long as the amp is correctly biased, both will work. I tried with 6V6 too but...meh...

    All that said, I'm not sure to understand how the output transformers can make such a difference, especially since I would bet they were made at the same factory (same markings on the lead, color codes, etc.) I suppose the power rating has an impact on tone (i.e. : Vibrolux have smaller OT than Bandmaster and that's part of why they sound different), the impedance too. But how does that explain that the PI tubes that suffered from oscillation with the first OT behave perfectly fine with the second one? Issues were worst with a long tail PI than with a cathodyne, but the 12ax7 was tremoloing with a cathodyne too.

    Anyway, I think I'm done with the power section and the PI. So I will now have to think about the preamp section. First thing I'll have to do is to inspect the grounding scheme, and clean the inputs, as the preamp is a bit noisy.
    The organ channel sounds good, so I'll leave it as is. The bass channel sounds terrible with a guitar, very dark and muddy, so I'll have to work on redesigning that part, but don't know exactly what to do with it.

    The first possibility is obsviously to design two gain stages and a tone stack, and voice them differently from the other channel. Each channel has its own tube, v3 being shared by both. So using something else than a 12ax7 would be possible too, as long as it doesn't draw more current for the heathers.

    But the amp has no obligation to have two channels, so it would be possible to make use of one dual triode and 3 controls to design something else. First idea I have is to implement a tremolo, but I would be curious if you have other suggestions? Or am I wrong to think the tremolo is feasible?
     
    HotBluePlates likes this.
  12. HotBluePlates

    HotBluePlates Member

    Messages:
    6,224
    Joined:
    Mar 27, 2017
    Folks who haven't designed a bunch of output sections (including me, at one point) totally misunderstand the OT loads they see used in amps & on data sheets.

    The primary impedance that's "best" for the tube depends on the supply voltage available, the amount of output power desired, and the distortion components allowed in the output section. It also depends on the designer's willingness to accept elevated screen current. And the specific OT selected (core size) depends on whether the designer wants to allow/avoid constricting bass response and OT core saturation for a certain power through-put.

    Point being, you can't determine the "optimum" OT primary impedance for a tube outside of knowing the supply voltage (and to some extent the desired power output). When you notice that certain "standard impedances" seem to repeatedly appear for certain tube types, it's because the power output & supply voltage is consistent across amp examples (and this implies class of operation is the same across those examples).

    Would it be fair to say the cause for your previous problems was never pinned down with certainty? If so, it could have been anything, from an issue with the OT (not very likely), to an oscillation (easiest to see with a scope & signal generator) or a poor solder connection (always a gremlin to watch out for).

    At this point, if the output section seems happy, probably best to leave it as-is and accept it's working well. :D

    The Organ channel is a 2-stage preamp lifted straight from a blackface Fender amp. It does add a 500pF across the 1st stage's plate load (that rolls off some treble), and has a Volume pot bright cap on a switch.

    The Bass channel is identical to the Organ channel, except that 1st stage plate load bypass cap is 20 times larger (sucks out much more highs/mids), and the tone stack has a change to the Mid resistor and cap (and adds a Fender-style "Deep" switch).

    If you yank that 0.01µF cap that's across the Bass channel's 1st stage 100kΩ plate load, your treble will come back.

    If you ditch a channel, tremolo is feasible. You could probably even keep both channels & add tremolo if you're willing to use a MOSFET as your trem oscillator (no tubes have to be re-allocated).

    But you might look into re-voicing one of the channels rather than swiping tubes for a different function. Maybe something like a 6G3 Deluxe preamp, where the higher plate & cathode resistances (you would use 3-3.3kΩ cathode resistor for a single tube section) change the operating point & tonality of the preamp.

    Loosing the 3-knob tone stack will greatly increase midrange & signal level. Or you could try looking into some different form of tone stack for this second channel.

    You might consider different cathode bypass caps in the 2nd channel; values as low as 1-2µF will give the stage all it's gain, except at the bottom octave of the guitar's range which will reduce bass congestion (especially when driven).

    It's also possible to implement reverb with just a single 12A_7, though you could probably have more options available from a good pedal than a shoe-horned reverb circuit.
     
    pdf64 likes this.
  13. HH1978

    HH1978 Member

    Messages:
    85
    Joined:
    Jun 9, 2019
    Thank you for you detailed reply! I think you're right on everything.

    (I could have pasted everything you wrote in that section). I admit I feel like shooting in the dark. I found so much conflicting informations on this... To the point that my only guideline was to find a set up that would look safe for the tubes and the OT. Supply voltage is known (about 360V for both the plates and screens), power output is not really a concern, anything from 15W to 40W would be suitable for the use of my bandmate. If there are good reliable sources to learn about output impedance, I'm certainly interested.

    Indeed, the cause is not known with certainty. It was suggested that the long tail inverter would not work with the available voltage, that's why I switched back to a cathodyne, but that doesn't exclude other possibilities as to why it didn't work. One thing I wonder, though, is why both the original cathodyne and the long tail were going into tremolo effect with a 12ax7, and the new cathodyne is not. I'm a newbie. Until recently, I was bringing my amps to the only (to my knowledge) tech in town. But twice I got an amp back in worst condition than before, so I decided to learn how to work things out myself. But when you go that way, you tend to learn from specific issues you encounter, to the detriment of a more global knowledge. I read what I could, but even then, I feel like I've got so much more to learn. The same goes for equipment, I don't have a scope, for instance, though I have a friend who's an electronical engineer, so I might have access to one if needed.

    But as you said, now it works and I'll leave it as is.

    Indeed, I did this yesterday. I was intrigued by the cap in parallel with the plate load, so I looked for informations on what it does, and once I found that, I decided to remove the cap.

    Thanks! I'll try the 6G3 set up. As for the tone stack, I could go for a James, or something very simple and effective like the one knob tone control found on the Trace Elliot Velocette/Gibson GA15, which uses a dual 250k pot. Depending on how it sounds, I might also switch the bypass cap for a smaller one.

    [​IMG]
     
    HotBluePlates and Tron Pesto like this.
  14. HotBluePlates

    HotBluePlates Member

    Messages:
    6,224
    Joined:
    Mar 27, 2017
    It's not "output impedance" but "optimum load" for the tubes.

    The output transformer simply uses transformer action to take a speaker load attached to the secondary, and "reflect" a higher load at the primary. So a transformer with a winding ratio of ~22.36 to 1 has an impedance ratio that is the square of this turns ratio, which is 500 to 1, and if an 8Ω speaker is attached to the secondary then a 8Ω * 500 = 4kΩ impedance is presented at the primary. This transformer gets marked "4kΩ primary, 8Ω secondary."

    The designer selects a loading that provides the performance they want, within the limitations imposed by the tube. There are some simple tricks to employ for figuring the output stage, but the most important thing is that you have a death-grip on Ohm's Law. That's because we usually don't need 4-decimal-place precision, and a first pass at an output stage design can use an "idealized tube" and pretend the load is a resistor. You then use Ohm's Law, and your understanding of peak, average, and RMS current, voltage & power to calculate power delivered into the "load resistor."

    Later, you can refine your calculations by noting how the real tube won't perform quite as well as the idealized tube, or how the power supply will deviate from your idealized assumption. In almost all cases, this just means you will wind up with less clean output power than you first calculated/predicted.

    • See the "Power-amp/Output Stage" section on the Valve Wizard site.
    • Know everything in Modules 1 & 2 (Direct Current & Alternating Current) of NEETS.
    • Kuehnel's Power Amps book probably has everything you need; I haven't read it but have read other books from him and they have good insights.
    • Radiotron Designer's Handbook 4th Edition Chapter 13 covers audio power amplifiers; the drawback is it is "just the facts" and if you're like me you might not see just how simple designing the output stage is without someone to talk you through it and prevent you from over-complicating the subject.
    • There are gobs of tube books at Tubebooks.org
    You're best served by learning output stage design from a source that's not bound up with guitar amplifiers (and giving you options for "TOANZ!"). It's just volts, amperes, ohms & watts, and knowing how to convert from peak to RMS or average values.

    You will benefit by seeing the + terminal of a filter capacitor as "ground" as far as a.c. is concerned. That's because it capacitor is a dead-short from B+ to ground for audio signals (well, 100Ω of reactance for a 20µF capacitor with an 80Hz signal).

    The capacitor across the plate load is "the same" as a capacitor from plate-to-ground, and reduces high frequencies at that stage.

    (You could also look at the capacitor's reactance as being in parallel with the plate load resistor, and effectively making the load resistance smaller as frequency rises. This reduces gain of a voltage amplifier stage, and can be proven with a couple of mathematical models.)​

    One channel of the 6G3 Deluxe has the same kind of cap-across-plate-load trick to make the channel a little less-bright than the other channel. I own two early-60's Deluxe amps and can tell you it's not too-extreme of a roll-off. It's not immediately obvious, but the 3000pF interacts with a lower total resistance than the 220kΩ, and implies a -3dB point of ~1100Hz.

    Note that "dual 250kΩ pot" has dissimilar sections.

    The upper section is audio taper and takes the place of the Treble pot in a Fender tone stack. The lower section is linear taper. At very low settings it acts like a Mid control in a Fender tone stack. But as you turn it higher, it decouples the tone stack from ground, acts like an Allen Amps Raw control, and increases overall signal level as it causes the tone stack to completely eliminate the mid-scoop inherent in that stack.

    The "Bass control" is replaced with a fixed resistor set higher than a Fender tone stack Bass control could ever be turned up to. Fun looking circuit, but the special pot may be a little harder to source. There's lots of room to model this in the Duncan Tone Stack Calculator, and tweak the circuit to try to get what you want.
     
    pdf64 likes this.
  15. HH1978

    HH1978 Member

    Messages:
    85
    Joined:
    Jun 9, 2019
    Thank you once again for the time you took to reply, and the loads of useful explanations and references! I was thinking about ordering Kuehnel's book, along with the preamps counterpart, now I'm sure I will. The NEETS link is awesome too! And actually, I downloaded RDH4 yesterday after noting that you decribed it as "bedrock" in another thread :)

    Actually, the pot is listed as "250k Lin dual gang" in the service manual, so I think the A and B letters on the schematic are just refering to the first and the second sections, but I could be wrong. I have a Velocette, so I should be able to check the tapers to confirm. If both sections are linear, the pot should be easy to source and I'll give it a try.
     
  16. HotBluePlates

    HotBluePlates Member

    Messages:
    6,224
    Joined:
    Mar 27, 2017
    I stand corrected! Yep, dual-linear so shouldn't be hard to find. I believe you're correct about the A & B designations referring to which section is used.
     

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