OK, I put a 68 ohm 5W resistor in the mains lead to the switch which brought ac down to 220v into the transformer. B+ still a bit high at 260v, but a long way better than it was. Transformer is running cooler too.
Sound has improved as well, better bass control. However still lot less clean across the range than my ECL86 SE. Gotta put a scope on it and see what's happening.
Traced the PCB against Ian444's schematic and it's pretty much the same, except the cathode resistors are 270+270 whereas Ian's were 180+180.
The power supply design is crude: 220v/110v transformer, a full wave voltage doubler for the HT allows the makers to use the same transformer for 110 and 220v mains by swapping winding connections. There are no balancing resistors across the capacitive split in the doubler, and no capacitors across the 1N4007 diodes. Definitely potential for improvement there.
The PCB assembly looks more like something a first-time kit builder would have done. Layout problems abound: obviously the board was laid out without checking the overlap of the tube sockets on surrounding components, meaning some components are mounted on one side and some on the other as they won't fit in their original spots because the socket is in the way. The grid stoppers (yes, they are there on mine) are BIG (no need) and are mounted too far away from the socket to really do their job. Easy fix (lift the grid pin of the socket from the PCB and install a small (maybe SMD) 10k resistor, strap out the old one). Will it help? Dunno
I think I will lay out my own hardwired PP on another chassis using these tubes and OPTs to settle on a design. If I find a good one, I'll transfer it to the Mini's chassis and ditch the PCB. Otherwise there are some good EL84 PP designs that could go in.
Gary
Sound has improved as well, better bass control. However still lot less clean across the range than my ECL86 SE. Gotta put a scope on it and see what's happening.
Traced the PCB against Ian444's schematic and it's pretty much the same, except the cathode resistors are 270+270 whereas Ian's were 180+180.
The power supply design is crude: 220v/110v transformer, a full wave voltage doubler for the HT allows the makers to use the same transformer for 110 and 220v mains by swapping winding connections. There are no balancing resistors across the capacitive split in the doubler, and no capacitors across the 1N4007 diodes. Definitely potential for improvement there.
The PCB assembly looks more like something a first-time kit builder would have done. Layout problems abound: obviously the board was laid out without checking the overlap of the tube sockets on surrounding components, meaning some components are mounted on one side and some on the other as they won't fit in their original spots because the socket is in the way. The grid stoppers (yes, they are there on mine) are BIG (no need) and are mounted too far away from the socket to really do their job. Easy fix (lift the grid pin of the socket from the PCB and install a small (maybe SMD) 10k resistor, strap out the old one). Will it help? Dunno
I think I will lay out my own hardwired PP on another chassis using these tubes and OPTs to settle on a design. If I find a good one, I'll transfer it to the Mini's chassis and ditch the PCB. Otherwise there are some good EL84 PP designs that could go in.
Gary
Yes it did, by leaps and bounds, but it is not finished yet. Great detail and deep bass, very happy with it, but I'm a relative newbie so keep that in mind... I'm thinking the OPT's are maybe half-decent for it to sound this good.
Ian.
Now off to see the Bondi Cigars at Ballina RSL. Happy New Year everyone!
Don't know if my procedure was the best having never worked on tubes before, so I used TrueRTA spectrum analyzer in a free run mode, and adjusted them for minimum harmonic distortion into a 7 Ohm load. It was only the second that was effected. Big improvement. I then found out you are "susposed" to measure the voltage on the output grids. Way off. If I balanced them, I was over 5% second harmonic, not .5. Sticking with the analyzer method. Thinking about it, it should be adjusting for the splitter, outputs, and any imbalance in the OPT. At least by what I have learned.
With all the 60Hz and harmonics, one would think adjusting the filament reference would make a difference. On the analyzer, it did not at all. So I adjusted them for midpoint. ( they were not) I suspect the rest of the noise is so bad, this in not a factor yet. I will revisit that when I ground the chassis. ( or build a larger chassis with some room to play. Or, I may just build regulated DC for them. The choice is voltage regulated or current regulated. I have seen arguments for each.
It looks like a long-tailed with CCS is a preferred topology, but without an input buffer stage, where would you bring the feedback to?
Consensus is the OTA's are not that bad. 6N2's are, and the textgbook , or as we say in solid state, "datasheet topology" needs help. Looking at about 50 schematics I downloaded, the more respected amps have an input buffer accepting the gnf, a long-tail pair, followed by a diff driver into the output. Looking at the really good ones, B+ for each tube separated and regulated, DC filaments, CCS feeds, and the ziener clamps as mentioned in AudioExpress.
If I add grid isolation resistors, is the 1K as shown a good value? I can't fnd the reference, but I remember someone, maybe Manley, talking about a range from 100 to 5K effecting the "tube" sound. Higher sounding progressivly more like solid state implying the circuit parastatics were a primary cause.
From the frequency response I ran on the RTA, it could use a tad of compensation in the feedback.
With all the 60Hz and harmonics, one would think adjusting the filament reference would make a difference. On the analyzer, it did not at all. So I adjusted them for midpoint. ( they were not) I suspect the rest of the noise is so bad, this in not a factor yet. I will revisit that when I ground the chassis. ( or build a larger chassis with some room to play. Or, I may just build regulated DC for them. The choice is voltage regulated or current regulated. I have seen arguments for each.
It looks like a long-tailed with CCS is a preferred topology, but without an input buffer stage, where would you bring the feedback to?
Consensus is the OTA's are not that bad. 6N2's are, and the textgbook , or as we say in solid state, "datasheet topology" needs help. Looking at about 50 schematics I downloaded, the more respected amps have an input buffer accepting the gnf, a long-tail pair, followed by a diff driver into the output. Looking at the really good ones, B+ for each tube separated and regulated, DC filaments, CCS feeds, and the ziener clamps as mentioned in AudioExpress.
If I add grid isolation resistors, is the 1K as shown a good value? I can't fnd the reference, but I remember someone, maybe Manley, talking about a range from 100 to 5K effecting the "tube" sound. Higher sounding progressivly more like solid state implying the circuit parastatics were a primary cause.
From the frequency response I ran on the RTA, it could use a tad of compensation in the feedback.
Amazing little differences. Trying to understand the why's.
The schematic above, if I understand, converts the original splitter to a conventional amp stage followed by a cathodyne splitter. My notes say this is not a very linear circuit. ( Used in my HK A300 I see) What was the advantage or set of tradeoffs for this choice?
Caps across the cathode resistors. Reduces feedback effect at higher frequencies. OK, why would one want this?
OPT's do not have the taps to run in super-linear mode. If this was such a great advance, thank you Williamson and Hafler, why not? Just cost?
6N1 vs 6N2 splitters, 6P1 vs 6P14 vs EL84 outputs. Other than adjusting resistor values for the current, why is one selected over the other? For that matter other than tiny difference in gain, 95 vs 100, why does the 6AX7 "sound" better.
If one were to not use the gNFB, then converting the phase splitter topology is easier. But, would not the gain of the amp go up along with odd order harmonic distortion? Sonically, why would I want this?
I do remember the reference on the grid resistors that they should be soldered as close to the pin as possible.
CCS in the cathode vs regulated supplies? Why and which?
The schematic above, if I understand, converts the original splitter to a conventional amp stage followed by a cathodyne splitter. My notes say this is not a very linear circuit. ( Used in my HK A300 I see) What was the advantage or set of tradeoffs for this choice?
Caps across the cathode resistors. Reduces feedback effect at higher frequencies. OK, why would one want this?
OPT's do not have the taps to run in super-linear mode. If this was such a great advance, thank you Williamson and Hafler, why not? Just cost?
6N1 vs 6N2 splitters, 6P1 vs 6P14 vs EL84 outputs. Other than adjusting resistor values for the current, why is one selected over the other? For that matter other than tiny difference in gain, 95 vs 100, why does the 6AX7 "sound" better.
If one were to not use the gNFB, then converting the phase splitter topology is easier. But, would not the gain of the amp go up along with odd order harmonic distortion? Sonically, why would I want this?
I do remember the reference on the grid resistors that they should be soldered as close to the pin as possible.
CCS in the cathode vs regulated supplies? Why and which?
Lots of questions...I don't know the answers to most of them.
The paraphase inverter had to go because no-one else uses it or recommends it, so I tried using the front-end and splitter of the Suppo Audio cct, and that sounds a lot better to me (cleaner, clearer sound). I will listen for a week or so and then try the Red Light District design which is similar, then a LTP with CCS. I'm just tinkering, not in any hurry, inspiration and activity seems to come in intermittent bursts. My notes say that the cathodyne and the LTP are both good choices
There can be conflicting info re tube designs, each author has their own preferences, the important bit I think is to know the limitations of any particular cct so you can keep it in the SOAR (safe operating area?) so to speak in SS terms.
I've been looking at the following amps, they are not listed in any particular order:
SY's Red Light District amp - 12AT7 triode stage with cathodyne splitter. Output tubes with tied cathodes and CCS.
Eli Duttman's El Cheapo - 12AT7 LTP with CCS. Output tubes with tied cathodes, one resistor and one capacitor.
Gingertube's Baby Huey - 12AX7 LTP with CCS. Output tubes with individual cathode resistors and caps, then 2 x CCS and caps, and finally fixed bias (3 different modes depending on development stage). Interesting local feedback.
The Machine by Poindexter - LTP with very long tail (450V total). Fixed bias output stage.
OddWatt/DIY Paradise Simple EL84 - input stage: SRPP/various triode, output stage: self-splitting.
A few of these designs use quite high voltages across the input stage, some use a B- supply to achieve this.
It is applied to the grid of the 2nd section of the LTP (El Cheapo, Baby Huey).
With smaller amps using EL84/6P1/6V6/6AQ5 the voltage swing required to drive the output tubes is not that large, so a simple input/splitter stage can work well if its a good one. With a simple design, the effects of each stage will be easier to hear, making the Mengyue or Suppo amps great value to learn with IMO.
Sound quality/characteristics in the particular cct, price, availability, B+ available, heater current draw if power tranny is already chosen.
6P14/EL84/6BQ5 have higher mutual conductance (and higher heater current) than 6P1/6AQ5/6V6. The 6P1 has the lowest max plate voltage at 250V. I don't know what the differences in sound are but with different pinouts it's difficult to swap between some types. Also 6AQ5 and 6V6 tubes do not use a 9-pin base.
As far as the small triodes go, it seems they need to be biased in their "sweet spot" wherever that is, the combination of correct voltage and current flow to give the best sound in the circuit they have been placed in, is my best guess at this stage. The tubes I have do not often match the datasheet very closely, but some do. For example, measure the cathode voltage and plate voltage of your 6AX7 or 6N1/6N2 and see if it corresponds to the datasheet under quiescent conditions. Maybe its just me? Same for your output tubes, what do you find? In the schematic I posted above, the 6N1P is getting barely a sniff of current and the B+ I think is way too low. The fact that it sounds good to me is surprising, but I've seen it happen before. For me its just trial and errror, if I think it needs more current, then I give it more and listen. Not very scientific. Sometimes more current does not help, but a different tube type will. When you dropped in the 6AX7, and it sounded better, you can look at the new bias point/operating area for the 6AX7 and maybe get some clues. I have not had any joy with 6N2 so far, but 12AT7 has been good.
The paraphase inverter had to go because no-one else uses it or recommends it, so I tried using the front-end and splitter of the Suppo Audio cct, and that sounds a lot better to me (cleaner, clearer sound). I will listen for a week or so and then try the Red Light District design which is similar, then a LTP with CCS. I'm just tinkering, not in any hurry, inspiration and activity seems to come in intermittent bursts. My notes say that the cathodyne and the LTP are both good choices
There can be conflicting info re tube designs, each author has their own preferences, the important bit I think is to know the limitations of any particular cct so you can keep it in the SOAR (safe operating area?) so to speak in SS terms.I've been looking at the following amps, they are not listed in any particular order:
SY's Red Light District amp - 12AT7 triode stage with cathodyne splitter. Output tubes with tied cathodes and CCS.
Eli Duttman's El Cheapo - 12AT7 LTP with CCS. Output tubes with tied cathodes, one resistor and one capacitor.
Gingertube's Baby Huey - 12AX7 LTP with CCS. Output tubes with individual cathode resistors and caps, then 2 x CCS and caps, and finally fixed bias (3 different modes depending on development stage). Interesting local feedback.
The Machine by Poindexter - LTP with very long tail (450V total). Fixed bias output stage.
OddWatt/DIY Paradise Simple EL84 - input stage: SRPP/various triode, output stage: self-splitting.
A few of these designs use quite high voltages across the input stage, some use a B- supply to achieve this.
It is applied to the grid of the 2nd section of the LTP (El Cheapo, Baby Huey).
With smaller amps using EL84/6P1/6V6/6AQ5 the voltage swing required to drive the output tubes is not that large, so a simple input/splitter stage can work well if its a good one. With a simple design, the effects of each stage will be easier to hear, making the Mengyue or Suppo amps great value to learn with IMO.
Sound quality/characteristics in the particular cct, price, availability, B+ available, heater current draw if power tranny is already chosen.
6P14/EL84/6BQ5 have higher mutual conductance (and higher heater current) than 6P1/6AQ5/6V6. The 6P1 has the lowest max plate voltage at 250V. I don't know what the differences in sound are but with different pinouts it's difficult to swap between some types. Also 6AQ5 and 6V6 tubes do not use a 9-pin base.
As far as the small triodes go, it seems they need to be biased in their "sweet spot" wherever that is, the combination of correct voltage and current flow to give the best sound in the circuit they have been placed in, is my best guess at this stage. The tubes I have do not often match the datasheet very closely, but some do. For example, measure the cathode voltage and plate voltage of your 6AX7 or 6N1/6N2 and see if it corresponds to the datasheet under quiescent conditions. Maybe its just me? Same for your output tubes, what do you find? In the schematic I posted above, the 6N1P is getting barely a sniff of current and the B+ I think is way too low. The fact that it sounds good to me is surprising, but I've seen it happen before. For me its just trial and errror, if I think it needs more current, then I give it more and listen. Not very scientific. Sometimes more current does not help, but a different tube type will. When you dropped in the 6AX7, and it sounded better, you can look at the new bias point/operating area for the 6AX7 and maybe get some clues. I have not had any joy with 6N2 so far, but 12AT7 has been good.
Last edited:
Much thanks. Found a couple of them.
The "Red-Light-District" is an interesting implementation of a CCS, I think.
Let's see if I am gettiing this. If you put an independent ccs on each cathode, then you have pretty much forced equal current, so class A mode. Cuts your power which may not be a help when starting with 10W. But, if you use one ccs, as in the red light, then the current is free to shift between the tubes in AB.
The "Red-Light-District" is an interesting implementation of a CCS, I think.
Let's see if I am gettiing this. If you put an independent ccs on each cathode, then you have pretty much forced equal current, so class A mode. Cuts your power which may not be a help when starting with 10W. But, if you use one ccs, as in the red light, then the current is free to shift between the tubes in AB.
Going through my old AA copies, I ran across an idea I have not seen here. Putting a diode/ziener clamp to ground on the grids of the outputs. This was written up as the greatest invention since super-linear. As I have not seen it anywhere else, makes me wonder. Is this just a fix for inappropriate margin in the design?
There's no CCS in that circuit. The cathodes of the output tubes have an LED array, which form a constant voltage source, exactly the opposite of a CCS.
I went ahead an rebuilt the PCB in mine to your circuit, although I bypassed the cathode resistor on the input (no gain without it). Sound is a little better, but still some noticeable colouration particularly on female vocals.
However, I still haven't connected the feedback so that is probably part of the problem. However, the skimpy power supply circuit is one area that concerns me most.
I don't really want to do any more experimenting on the PCB, as it was already minus a couple of pads from the factory, and I lifted a couple more with almost no effort and a good soldering iron. You must have got lucky.
It's quite livable like it is, and I think I will experiment with the same tube types (have a bunch of spares coming) on another chassis.
If I hardwire this one, it will get EL84's instead.
Have you done any more toying with your amp?
Gary
My bad SY, I stated the opposite in a summary a page back (post 29), leading tvr astray. Once pointed out, it is obvious. Thanks for clearing that up. The RLD thread and your web article sure are a very interesting read, i.e. the balancing of the amp requirements during design stage to arrive at the best solutions for the design goals.
Looking at the B+'s used on the input stages of all the amp designs I mentioned in that post, I notice that nearly all of them use a total B+ on the input stage over 350V. It appears no matter what cct is built into the Meng Yue, the low B+ available for the input stage is a challenge.
The Meng uses a full-wave voltage doubler connected to the secondary of the power transformer. It may be possible to make a tripler to give around 370V with the addition of a cap and a diode off the doubler. I knocked up a copy of the amp's doubler with added half-wave tripler at work on a breadboard and the principle appears to work.
Gary, sorry to hear the improvement wasn't that great. Are you using 6N1P's or the original 6N2's? I have not found a lack of gain, if you like, I can run a sine wave through it and record some AC voltages. Another method of rebuilding an amp like this is to pull the PCB out and scrape all the tracks off it, then rebuild point to point. This was suggested to me by an experienced aussie amp builder but I've never tried it yet. I haven't done any more tinkering but will this weekend. I got oscillation when I connected feedback, or more correctly, when I added a 100pF cap in parallel with the 24K feedback resistor. I checked the phase with a scope. Left the feedback unconnected. I can't hear any colouration.
Ian.
Last edited:
My books from Morgan Jones arrived. Maybe I can quit with the hoof and mouth. I have the Meng and an HK A300 to play with and a TeleMatic, which I can try and reverse engineer. If the iron or chassis is good, I may make a preamp out of it. Antique box with totally modern guts.
As my NAD blew up, I hooked up the Meng with the 6AX7's. ( CD to passive switch box) Ya' know, it does not listen that bad. No bass, but listenable.
As my NAD blew up, I hooked up the Meng with the 6AX7's. ( CD to passive switch box) Ya' know, it does not listen that bad. No bass, but listenable.
I am using the original tubes, but as the printing is illegible on them they could be either type.
Gain is only slightly down, and only on the original circuit being run on excessive mains input which made it on the verge of oscillating I reckon. It will still reach clipping anyway so it's not a problem.
Did you run the input stage with cathode bypassing? Your diagram shows none, but without it I have very low gain (but very clean). The difference was that I was getting a max of 1W output for full drive from my computer on 1khz tone. After bypassing I get that output with the volume down to below 1/4 turn. My cathode resistor is 2k2 bypassed by 100uF and the cathode voltage is 0.76v.
I would love to scope things, but my cheap portable LCD scope won't stand anything over 50v from memory. And a daisy chain of clip leads from the probe to a capacitor to the amp doesn't sound like a safe way to measure anode signals. I should buy a 10x probe for it I suppose.
Gary
No Bass? Mine has always had plenty of bass. I think Ian said his did too.
Mine is at least on a par with my Pioneer 100W/ch s/s theater receiver.
From experience with unstable solid state amps, if you have parasitic oscillation at very high frequencies the audible symptom is lack of bass (and hot output transistors). Might pay to check. Reversing the phasing of one of the speakers can do it too but that should be obvious.
Gary
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.