Hi
Some months ago, I finally got my hands on an Audio Research Mkiii and I want to learn as much as possible about it. Its a fully balanced linestage, uses 8x 6922 tubes. Since it may need new tubes, I have some questions.
1. When buying tubes, how should they be matched, in pairs or quads?
2. What is the purpose of RV2 and RV3 in the schematic
3. Which current production tubes? I'm currently looking at Gold Lion 6922's of Tubedepot, or perhaps some Phillips ECG 6922's off an ebay seller. Due to the high number of tubes, I'm not really interested in NOS tubes like NOS Amperex 7308 (even if they're the best sounding), my upper budget would be the Gold Lion 6922's , they're currently about $75 each.
Attaching circuit diagram if it helps answering the tube matching question, hopefully its clear enough.
Looking at the schematic, looks like this is for a single channel?
Any ideas on what Q1-Q5 do? They're color coded with paint by ARC.
Thanks in advance!
Some months ago, I finally got my hands on an Audio Research Mkiii and I want to learn as much as possible about it. Its a fully balanced linestage, uses 8x 6922 tubes. Since it may need new tubes, I have some questions.
1. When buying tubes, how should they be matched, in pairs or quads?
2. What is the purpose of RV2 and RV3 in the schematic
3. Which current production tubes? I'm currently looking at Gold Lion 6922's of Tubedepot, or perhaps some Phillips ECG 6922's off an ebay seller. Due to the high number of tubes, I'm not really interested in NOS tubes like NOS Amperex 7308 (even if they're the best sounding), my upper budget would be the Gold Lion 6922's , they're currently about $75 each.
Attaching circuit diagram if it helps answering the tube matching question, hopefully its clear enough.
Looking at the schematic, looks like this is for a single channel?
Any ideas on what Q1-Q5 do? They're color coded with paint by ARC.
Thanks in advance!
Attachments
The fets are hand-selected current sources. Avoid touching the adjustments unless you have the service manual.
V1 and V3 should be a matched pair, since they are in parallel. Same for V2 and V4.
Ideally, V1-V4 should be a matched quad.
V1 and V3 should be a matched pair, since they are in parallel. Same for V2 and V4.
Ideally, V1-V4 should be a matched quad.
As long as the valves are of good quality, don't get sucked into paying for matched valves. In triodes, it makes little difference except costing you more cash.
RV2 and RV3 set up the bias point for the first two valves, (much like matching them).
6922 is an ECC88, 6n23P or 6n1p etc etc.
I have had best results from the 6n23P or 6n1p from Slovakia. A matched Quad from https://www.ebay.co.uk/itm/386965048863?_skw=6n23p&itmmeta=01JDFZ6JY1AP32FVENVG2CEW2E&hash=item5a18ea0e1f:g:RDIAAOSwGt1mLL2j&itmprp=enc:AQAJAAAA4HoV3kP08IDx+KZ9MfhVJKnbwyoqA0S64N++EVTRacXciYNRGKEQH1ARZwdXYlvx0R8f4U5mRkLAMzdZ4hHIB2DQRhAtjRuhtvzUjwBDdMGY4Bd0T8fnhJHCG+G/ZvyirKiAcv9l7Zkg+gRJTa0rxxxgavLKk/aB80JDsUJxRNvIVauGL4kLIlxd39G58HbJFraKhOfiOYO6gsQprfOSffUvRGbF+L6zIZ7zPrKseabJ2Rln6R2KreEjEEEpNdgZpSl/hYuhsLPALJVFUZORjumR553CAtiMj8/4A2b6oIsG|tkp:BFBMoq-a_-tk
will cost you around $10 At that price, you can afford to buy plenty and match them yourself.
Why pay more?
RV2 and RV3 set up the bias point for the first two valves, (much like matching them).
6922 is an ECC88, 6n23P or 6n1p etc etc.
I have had best results from the 6n23P or 6n1p from Slovakia. A matched Quad from https://www.ebay.co.uk/itm/386965048863?_skw=6n23p&itmmeta=01JDFZ6JY1AP32FVENVG2CEW2E&hash=item5a18ea0e1f:g:RDIAAOSwGt1mLL2j&itmprp=enc:AQAJAAAA4HoV3kP08IDx+KZ9MfhVJKnbwyoqA0S64N++EVTRacXciYNRGKEQH1ARZwdXYlvx0R8f4U5mRkLAMzdZ4hHIB2DQRhAtjRuhtvzUjwBDdMGY4Bd0T8fnhJHCG+G/ZvyirKiAcv9l7Zkg+gRJTa0rxxxgavLKk/aB80JDsUJxRNvIVauGL4kLIlxd39G58HbJFraKhOfiOYO6gsQprfOSffUvRGbF+L6zIZ7zPrKseabJ2Rln6R2KreEjEEEpNdgZpSl/hYuhsLPALJVFUZORjumR553CAtiMj8/4A2b6oIsG|tkp:BFBMoq-a_-tk
will cost you around $10 At that price, you can afford to buy plenty and match them yourself.
Why pay more?
But the grids are not. One pair of grids is driven by the input the other by feedback.
6922 = 6DJ8 = 6N23P, heaters need 300mA each. 6N1P has double the heater current draw (600mA @ 6.3v) and 8 of those pull 4.8A at 6.3v, where 8x 6DJ8 only pull 2.4A.
ok, some 6922 pull 335mA. but still.
Hi All
Thanks for the feedback so far, very much appreciated.
Would the bias point need adjusting based on the tubes individual characteristics?
The power supply section of the LS5 Mkiii seems to show that the heaters are powered with a pair of LM317's, based on the information @dubabub posted, with the 6N1P current at 600mA, then that would mean that the 6N1P would not be suitable selection it seems?
Thanks for the feedback so far, very much appreciated.
Would the bias point need adjusting based on the tubes individual characteristics?
The power supply section of the LS5 Mkiii seems to show that the heaters are powered with a pair of LM317's, based on the information @dubabub posted, with the 6N1P current at 600mA, then that would mean that the 6N1P would not be suitable selection it seems?
Attachments
Hi
Why is the first thing you want to do is to change all the tubes? Does the preamp work? Is it noisy? Are the pots scratchy?
According to manufacturer's data sheets, a tube is not considered to be anything close to "dead" until the cathode emission falls below half of its rating. That takes decades with moist tubes. The reality is that tubes will last for generations provided they are not mechanically upset. Most hifi gear lead sedate lives sitting on shelves. Best to keep the tubes out of the vibration fields of subs and other speakers.
Audio Research made many amps and preamps with balanced circuitry and cross-coupled feedback - this is another of the same. Unless you have a repair manual that tells you how to adjust the two current sources, it is best to leave them alone. In the factory, they would have set things using a THD analyser, at least for first units, and possibly come up with a simpler, quicker procedure for production. However, if production volume is low the THD method is workable, especially once automatic integrated units like from Amber came along.
Pots need occasional spray of lube. The thicker stuff like Blue Stuff cleans and lubricates. DO NOT use WD-40 or similar as those chemicals will eat the carbon track of the pot.
If the preamp has not been used for a long time, the components most in need of attention will be electrolytic caps. These require regular application of voltage to stay healthy. Their service life for best performance is about 14-years.
Carbon resistors tend to be noisy and also exhibit voltage-dependent distortion. Some manufacturers use higher power rated types to try to reduce thermal noise and improve reliability. For some, the unique distortions of carbon resistors are part of the magic; for others, it is a liability and they replace carbon with metal-film or even metal-foil ($$$).
There is no need to spend any more money than you have to on these things unless doing so makes you happy.
Why is the first thing you want to do is to change all the tubes? Does the preamp work? Is it noisy? Are the pots scratchy?
According to manufacturer's data sheets, a tube is not considered to be anything close to "dead" until the cathode emission falls below half of its rating. That takes decades with moist tubes. The reality is that tubes will last for generations provided they are not mechanically upset. Most hifi gear lead sedate lives sitting on shelves. Best to keep the tubes out of the vibration fields of subs and other speakers.
Audio Research made many amps and preamps with balanced circuitry and cross-coupled feedback - this is another of the same. Unless you have a repair manual that tells you how to adjust the two current sources, it is best to leave them alone. In the factory, they would have set things using a THD analyser, at least for first units, and possibly come up with a simpler, quicker procedure for production. However, if production volume is low the THD method is workable, especially once automatic integrated units like from Amber came along.
Pots need occasional spray of lube. The thicker stuff like Blue Stuff cleans and lubricates. DO NOT use WD-40 or similar as those chemicals will eat the carbon track of the pot.
If the preamp has not been used for a long time, the components most in need of attention will be electrolytic caps. These require regular application of voltage to stay healthy. Their service life for best performance is about 14-years.
Carbon resistors tend to be noisy and also exhibit voltage-dependent distortion. Some manufacturers use higher power rated types to try to reduce thermal noise and improve reliability. For some, the unique distortions of carbon resistors are part of the magic; for others, it is a liability and they replace carbon with metal-film or even metal-foil ($$$).
There is no need to spend any more money than you have to on these things unless doing so makes you happy.
The preamp was purchased around 1996/7, has never been serviced. It was used a few times a week 4-5 hours at a time up until 1998. It then sat relatively dormant (used perhaps 2-3 times a year) for the next 25 years. When I finally went to pick it up and bring it back with me to the US, I got a chance to listen to it in its original system for the last time. There were issues, one channel didn't work, not sure whether this was source, preamp or amp related, but we left the system powered up playing at low volume, and the dead channel came back and continued to work. Volume control was definitely scratchy.
Does the preamp work? Yes it does
Is it noisy? No its not
Are the pots scratchy? Yes they were, but not anymore.
Why do I want to change all the tubes? To experiment with a different sound. I feel that it lacks top end and the bass could be better. The mids are still very good. I'm comparing it to my DIY Tubes4hifi SP-14 and I'm wondering whether the LS5 Mkiii could be better, right now, it isn't better, but different.
It still has the original Sovtek tubes, but I've read that there are better tubes out there. The first thing I want to do is get it back to Audio Research to have it thoroughly checked and have the electrolytics replaced. That's at least $500, excluding shipping and anything else they feel they need to do plus labor. If the result is that if tubes need replacing, I'd like to know my options beforehand.
I haven't made any adjustments to RV2 and RV3, I wanted to learn about this preamps circuit and figured the smart folks on this forum (who may have direct experience with this model), given a schematic, might be able to help me understand it better.
If by changing tubes you can hear a difference then the tubes are faulty. All mentions of "high" and "low" is meaningless as this amp will amplify all frequencys alike. That's the function of a well designed amp.
If you want more or less of high and low get a equalizer and use that between the AR and the power stage.
If you want more or less of high and low get a equalizer and use that between the AR and the power stage.
Hi
ALL FRESH NEW TUBES WILL SOUND BRIGHTER THAN WHAT IS IN THERE NOW.
Do not confuse that with "the new ones are better" or "the old ones are dead".
New tubes need to be broken in for about 100 hours and then they will be past their infant tone and be on their tone plateau. During the break-in, the tube loses a bit of gain, harshness mistaken as brightness, and becomes more how the manufacturer has rated the device. On the tone plateau, gain is lost at an indiscernible rate and will be stable for decades. Any distortion "promises" you see in application notes are done with broken-in tubes. Also, it behooves one to evaluate tubes of different brands on the technical merits rather than on nationalistic prejudice. This may save you a lot of $$
Flat frequency response and low-THD for some people will sound "flat" and "uninteresting". Maybe you are just used to a higher THD or a different THD profile? Everyone has their own preferences.
The front end of the AR pre is weird inasmuch as there are two balanced diff amps across the signal phases, but they are connected to want to be two diff amps within each phase. AR chose to make the cross-phase diffs the true diff amps - prioritising signal balance over THD reduction.
ALL FRESH NEW TUBES WILL SOUND BRIGHTER THAN WHAT IS IN THERE NOW.
Do not confuse that with "the new ones are better" or "the old ones are dead".
New tubes need to be broken in for about 100 hours and then they will be past their infant tone and be on their tone plateau. During the break-in, the tube loses a bit of gain, harshness mistaken as brightness, and becomes more how the manufacturer has rated the device. On the tone plateau, gain is lost at an indiscernible rate and will be stable for decades. Any distortion "promises" you see in application notes are done with broken-in tubes. Also, it behooves one to evaluate tubes of different brands on the technical merits rather than on nationalistic prejudice. This may save you a lot of $$
Flat frequency response and low-THD for some people will sound "flat" and "uninteresting". Maybe you are just used to a higher THD or a different THD profile? Everyone has their own preferences.
The front end of the AR pre is weird inasmuch as there are two balanced diff amps across the signal phases, but they are connected to want to be two diff amps within each phase. AR chose to make the cross-phase diffs the true diff amps - prioritising signal balance over THD reduction.
Hi
I was curious about the crossed feedback loops in this preamp, so did a sim in LTspice. I did not have 6922 models and just used 12AT7 to see the circuit function. I used 8m8 current sources for the two diff amps and the paraphase stage as that got the voltages to be around what the schematic showed. At first I had current sources for the output CFs but the negative phase would not work. I changed those to a jfet running at 1m6 each. Things you can get away with in spice 🙂
With 1Vpk in to both sides, there is 4V5 pk at each output at 0.05%THD at 1kHz. With the 12AT7s frequency response goes out to 3MHz but phase starts to drop at 100kHz.
THD jumps to 3.7% each side if the feedback loops are disabled (R15,16=0) and the output jumps to 18Vpk and is still fairly symmetrical.
Reconnecting the positive NFB path (R16=15k) restores the positive output to 0.05% and the negative side to 0.1%, both at a slightly elevated 7v4.
Shorting the negative input with 1Vpk to the positive input produces symmetric 2v3 outputs at 0.026% each. Reversing this produces identical results.
Like most balanced circuits, they really want to have two input signals.
I was curious about the crossed feedback loops in this preamp, so did a sim in LTspice. I did not have 6922 models and just used 12AT7 to see the circuit function. I used 8m8 current sources for the two diff amps and the paraphase stage as that got the voltages to be around what the schematic showed. At first I had current sources for the output CFs but the negative phase would not work. I changed those to a jfet running at 1m6 each. Things you can get away with in spice 🙂
With 1Vpk in to both sides, there is 4V5 pk at each output at 0.05%THD at 1kHz. With the 12AT7s frequency response goes out to 3MHz but phase starts to drop at 100kHz.
THD jumps to 3.7% each side if the feedback loops are disabled (R15,16=0) and the output jumps to 18Vpk and is still fairly symmetrical.
Reconnecting the positive NFB path (R16=15k) restores the positive output to 0.05% and the negative side to 0.1%, both at a slightly elevated 7v4.
Shorting the negative input with 1Vpk to the positive input produces symmetric 2v3 outputs at 0.026% each. Reversing this produces identical results.
Like most balanced circuits, they really want to have two input signals.
Hi
One other service item with old gear is cold solder joints. Fortunately, older equipment was often hand soldered and the connections are large compared to later automated soldering, and gigantic compared with surface-mount solder connections. Solder is eroded by electron flow, so having a large connection to begin with favours a long problem-free duty before the connections need to be redone.
Connections susceptible to heat or mechanical stress, for example on tube sockets, are the ones to look at first.
You cannot simply add new solder to old. You have to remove the old solder and redo it, but to do this, you in fact add new solder to the old to assist getting the old solder to flow for removal with a solder sucker. Then resolder the connection with new solder.
Use leaded solder (like Kester 44) if you want connections that you can visually tell are good. They should be shiny and the wetting should be concave. You can use the "no clean" 245 solder from Kester, but the finished joint will look dull. NEVER use lead-free or silver solder . With the former it is impossible to tell if the joint is good simply by looking at it. Silver solder requires tons of heat and you may damage components. Besides, authorities I respect tell me that using "a bit of silver" is worse than having no silver at all, that in fact you must have silver wire, silver leads, silver windings, etc to make proper use of silver solder. After a while, all the silver tarnishes and looks terrible.
One other service item with old gear is cold solder joints. Fortunately, older equipment was often hand soldered and the connections are large compared to later automated soldering, and gigantic compared with surface-mount solder connections. Solder is eroded by electron flow, so having a large connection to begin with favours a long problem-free duty before the connections need to be redone.
Connections susceptible to heat or mechanical stress, for example on tube sockets, are the ones to look at first.
You cannot simply add new solder to old. You have to remove the old solder and redo it, but to do this, you in fact add new solder to the old to assist getting the old solder to flow for removal with a solder sucker. Then resolder the connection with new solder.
Use leaded solder (like Kester 44) if you want connections that you can visually tell are good. They should be shiny and the wetting should be concave. You can use the "no clean" 245 solder from Kester, but the finished joint will look dull. NEVER use lead-free or silver solder . With the former it is impossible to tell if the joint is good simply by looking at it. Silver solder requires tons of heat and you may damage components. Besides, authorities I respect tell me that using "a bit of silver" is worse than having no silver at all, that in fact you must have silver wire, silver leads, silver windings, etc to make proper use of silver solder. After a while, all the silver tarnishes and looks terrible.
Hi Nauta
Thank-you very much for taking the time to respond. I'll admit that what you have simulated here is currently a little too advanced for me to understand right now. I'll continue to learn more about tube circuits in general.
So to summarize, it seems like its best to run this preamp in balance mode? It does have a switch to flip between balanced and single ended mode.
Thanks for the tips on working on old gear. I already do have a roll of Kester 44, nice stuff to work with, no issues with it yet. I used to have a roll of lead free solder, and yeah, its definitely more difficult to work with. Perhaps I might get some backlash on this, but I also have a roll of Cardas Quad Eutectic solder which I haven't used yet, was curious on how that would perform.
Are we saying that all tubes of the same type will sound the same in the same preamp? Or, they could still sound different because tubes of different brands may behave differently in the same circuit.
What should I look out for when choosing tubes from Gold Lion 6922 vs "NOS" Phillips ECG 6922? These are currently on my radar, the Gold Lions are twice the price of the Phillips.
Hi
The Kester 44 is standard tin-lead solder with a 60-40 ratio. When you heat it, it goes from solid to a plastic state then to a liquid. Eutectic solder has a slightly different ratio of 63-37 so it goes directly from a solid to a liquid when heated. Maybe the 44 is available in that ratio, too, but I never saw it or was aware of it. Both of these are rosin-core. Do not use separate rosin as you will make a mess. Let the residue harden before chipping it off. Do not use chemical cleaners as you really have to flush the board to get it fully clean and if it is not clean, the chemicals left behind will eat the board and create stray pathways between traces.
As far as the same type of tubes go, presumably to make a 6922, say, the geometry has to be very specific to achieve the characteristics expected. If multiple manufacturers are allowed to make the tube, their "6922" has to be compatible with every other "6922" so they will all work properly in equipment designed for it. So, to me, this suggests that once all the various samples are burned in, there should be no difference between their performance. If you get a 6922 from one supplier and it does not perform the same way, assuming it is not defective, then it is likely that manufacturer actually made a different tube and marked it as the desired tube.
When Mike Matthews of Electroharmonix bought all the famous brand names and either bought a Chinese factory or had enough $$ sway to get them to make what he wanted, he could basically mix and match heaters, cathodes, grids, screens, plates and bases and envelopes to create variations of tubes that never existed before. He could make getters different shapes in basically identical tubes and brand them how he wanted and charge appropriate cash for the cache. The getter is only used during manufacturing and is not an active part of the tube during operation, but their shape is used to identify "real" brand tubes..
Regarding using the balanced preamp in single-ended mode: THD will be a bit higher. The point of having balanced connections in an audio system even right to the speaker drive - maybe especially for speaker drive - is to eliminate the distortion of the signal through ground current disturbances,
The Kester 44 is standard tin-lead solder with a 60-40 ratio. When you heat it, it goes from solid to a plastic state then to a liquid. Eutectic solder has a slightly different ratio of 63-37 so it goes directly from a solid to a liquid when heated. Maybe the 44 is available in that ratio, too, but I never saw it or was aware of it. Both of these are rosin-core. Do not use separate rosin as you will make a mess. Let the residue harden before chipping it off. Do not use chemical cleaners as you really have to flush the board to get it fully clean and if it is not clean, the chemicals left behind will eat the board and create stray pathways between traces.
As far as the same type of tubes go, presumably to make a 6922, say, the geometry has to be very specific to achieve the characteristics expected. If multiple manufacturers are allowed to make the tube, their "6922" has to be compatible with every other "6922" so they will all work properly in equipment designed for it. So, to me, this suggests that once all the various samples are burned in, there should be no difference between their performance. If you get a 6922 from one supplier and it does not perform the same way, assuming it is not defective, then it is likely that manufacturer actually made a different tube and marked it as the desired tube.
When Mike Matthews of Electroharmonix bought all the famous brand names and either bought a Chinese factory or had enough $$ sway to get them to make what he wanted, he could basically mix and match heaters, cathodes, grids, screens, plates and bases and envelopes to create variations of tubes that never existed before. He could make getters different shapes in basically identical tubes and brand them how he wanted and charge appropriate cash for the cache. The getter is only used during manufacturing and is not an active part of the tube during operation, but their shape is used to identify "real" brand tubes..
Regarding using the balanced preamp in single-ended mode: THD will be a bit higher. The point of having balanced connections in an audio system even right to the speaker drive - maybe especially for speaker drive - is to eliminate the distortion of the signal through ground current disturbances,
Hi
A clarification to post-14:
With 1V peak signal on both inputs producing 4v5 peak from each output, you could look at it as 2V peak input producing 9V peak output. The gain from each input to its own output is 4.5, and so is the total input measured across the inputs compared to both outputs measured across both outputs.
With one input shorted, there is 1Vpk to the unshorted input making 2v3 on each output, so 1V in and 4v6 out. The 2v3 was a rounding, so really we still have the 4.5 gain measuring across both outputs and both inputs.
If you use just one input phase and only one output phase, the gain is 2.25, half of the differential signal gain.
From input to output, the circuit is one. Its use of differential amplifiers intimately ties the two signal paths together and we achieve the performance described above 🙂
There are "balanced" amps and preamps that are built as independent amplifying paths and they do not exhibit these characteristics. Instead, a signal applied to only one input produces only one signal on the related output.
A clarification to post-14:
With 1V peak signal on both inputs producing 4v5 peak from each output, you could look at it as 2V peak input producing 9V peak output. The gain from each input to its own output is 4.5, and so is the total input measured across the inputs compared to both outputs measured across both outputs.
With one input shorted, there is 1Vpk to the unshorted input making 2v3 on each output, so 1V in and 4v6 out. The 2v3 was a rounding, so really we still have the 4.5 gain measuring across both outputs and both inputs.
If you use just one input phase and only one output phase, the gain is 2.25, half of the differential signal gain.
From input to output, the circuit is one. Its use of differential amplifiers intimately ties the two signal paths together and we achieve the performance described above 🙂
There are "balanced" amps and preamps that are built as independent amplifying paths and they do not exhibit these characteristics. Instead, a signal applied to only one input produces only one signal on the related output.