Luckiestmanalive Posted October 7, 2023 Author Posted October 7, 2023 Took me a few weeks but number 3 on the to-do list (change from Pansonic to Auricaps) is done (see pic below). It's been a couple of weeks since it was done and it wasn't a night and day difference. However, I have had some correspondence with the guy in Japan who blogged his experience restoring an A3600 (see this episode of iruchan blog). He posted a picture of the configuration he was working from and I decided yesterday to increase the cathode resistors (R113-116) for the phase inverter from 9000 to 10,000 ohms to see what change it makes. According to the ampbooks LTP calculator, balancing the 6SN7 using the existing plate resistor requires a resistor somewhere between 9-10Kohm. What I found is that this provides about a 5V boost to the plate voltage of both the phase inverter and input tubes and the cathode voltage of the phase inverter (first and second stages are directly coupled). While I was evaluating the consequences of the change on vtadiy.com I looked at both 6CG7 and 6GU7 tubes and noticed the 3rd order distortion is much lower using 6GU7s in the current configuration (like 0.01% vs 0.6%). So I've just swapped back to 6GU7s and had a quick evaluation. This change is a very noticeable improvement. More realistic timbre, separation and soundstage cohesion. I think I'll be sticking with this configuration for a while. Last on my to-do list is mucking about with negative feedback but I'm in no hurry... 1
Luckiestmanalive Posted October 16, 2023 Author Posted October 16, 2023 (edited) I've been tube rolling ever since I got this amplifier and I've taken the opportunity to measure the voltages using the handy test points on the A3600s PCB. I have built up a small tube collection consisting of about a dozen input signal tubes (6AQ8/ECC85s), and same again of 6CG7/6FQ7s and 6GU7s (driver tubes). I created a spreadsheet for different combos to isolate the best tubes of each type. While I don't have a tube tester, I have enough combos to establish which tubes are balanced. It is no wonder that people can hear the difference between tubes, especially in a PP amp with direct coupling between the input and driver stages. Just by the voltages alone, there is a significant variation around those indicated in the schematic, which means the tubes can be operating outside the area the designer intended, increasing distortion and reducing sound quality. In particular, the A3600 design intends the 6AQ8 to operate with a B+ of 260V, a plate load resistor of 270Kohm, a plate voltage of 80V, and have a cathode voltage of 1.6V from a cathode resistance of 2.3Kohm bypassed with a 100uf cap. From my dozen samples of twin triodes, I've taken readings on the plate anywhere between 75-84V (with 9Kohm driver cathode resistor) and 80-90V (with 10Kohm driver cathode resistor). According to vtadiy.com's universal loadline calculator, third harmonic distortion is minimal (<0.1%) if the plate voltage less cathode voltage lies somewhere between 78-87V. After taking a few readings of different combos of signal and driver tubes, I've narrowed down my stash of 6AQ8s to four that have plate voltages within 1V on each side within an optimal zone (82-88V). However, I am still perplexed by a discrepancy between the model in vtadiy.com for the 6AQ8 and my experience with cathode voltages, which vary from 1.4-1.55V when the model says it should be 1.6V or higher, a discrepancy I can't explain... I found one 6AQ8 that is nearly perfectly matched on both sides and, with it in place, I went about swapping the driver tubes in left and right positions to find ones that matched, left and right (not while it was on!). Being an LTP driver stage, the signal goes through one triode and then the other so, with the benefit of gain from the first, the voltage is lower for the second side. The plate voltages should be 350V and 340V, respectively, but I found the variation from the tubes I had (which differ from the original 6240Gs) meant plate voltage is anything from 310V to 350V. However, I don't think this voltage matters as much as the shared cathode current source from the single cathode resistor (as long as they are balanced left and right). The 9Kohm or 10Kohm cathode resistor results in ~90V or ~100V so Ohms law suggests ~10mA shared between the two triodes. According to vtadiy.com, with 6GU7 tubes, third harmonic distortion is minimal (<0.03%) if the QOP is above 170V (up to 290V!). With 6CG7 tubes, third harmonic distortion is below 0.4% if the QOP is between a range between 162-179V. I have a good ear, perhaps helped by my training as a classical pianist, so I can hear that the amplifier sounds better with 6GU7 tubes and that is the combo I'm using now. I've found one pair of 6CG7/6FQ7 and 6GU7 tubes that at least match on the high and low sides within a few volts. The balanced combo of tubes that roughly match the difference in voltages left and right result in roughly the following voltages (in case anyone wants to compare). With B2 of anywhere between 470-480V (470V in schematic and variation caused by differences in the line voltage): 6AQ8: B3 and B4 ~260V, Plate voltage 85.5V/84.6V (1V less with 6CG7s), Cathode voltage 1.47/1.48V (1.5/1.51V with 6CG7s) 6GU7: B2 less cathode voltage 381/382V (L/R), Plate voltages 333V & 322V/333V & 323V (L/R), Cathode voltages 98V/97V (L/R) 6CG7: B2 less cathode voltage 486/487 (L/R), Plate voltages 330V & 321V/333V & 321V (L/R), Cathode voltages 94/93V (L/R) My two matched pairs of KT120s experience a plate voltage of close to 500V, ultralinear (44%) taps to screens (without screen resistors), fixed biased to 75mA, reactive load between 4000-6000ohms, resulting in only third harmonic (due to PP cancellation of even harmonics) below 0.12%, according to vtadiy.com. Edited October 16, 2023 by Luckiestmanalive 1
Luckiestmanalive Posted October 16, 2023 Author Posted October 16, 2023 Thanks, C! It has certainly been a journey of discovery! I guess it's what happens if you choose to buy an amplifier that forces you to understand how they work after a lifetime of simply switching components in and out to try to improve the music! I have the community to thank for writing or videoing about their own journeys. The least I can do is do is return the favour, even if it isn't perfect!
couchy Posted October 16, 2023 Posted October 16, 2023 Hi Greg,I have a few luxman pieces which I love . I have the mb3045 monoblocks which use the same original 8045 valves. I have a spare pair of the obsolete tubes too. They sound brilliant but I bet they could sound better with some updated parts. Not sure I would be up to the challenge like you . They sure did build the old Lux's well. Enjoy. 1 1
Luckiestmanalive Posted October 17, 2023 Author Posted October 17, 2023 (edited) Wow! You have the 3045s - I bet they sound fantastic! If you have good original tubes (both 6240Gs and 8045Gs) then I doubt updating the parts would improve it. I've only changed mine because the amp needed adjustments to work well with other driver and power tubes. But I'm well pleased with the result and I've learned so much as a result - there being no official circuit but the original one for the original tubes, I can make adjustments as I choose. I'm not going for resale value. This is my amplifier, designed by me (in part). Edited October 17, 2023 by Luckiestmanalive 1
couchy Posted October 17, 2023 Posted October 17, 2023 I mean things like recapping old parts that might be out of spec. The amps have never been serviced from my knowledge. 1
Luckiestmanalive Posted October 17, 2023 Author Posted October 17, 2023 Yeah, I changed the electrolytics in the power supply as a precaution but I measured them after I pulled them out and they were a little high but still within spec. While this doesn't mean they were good, there were no other signs they were going bad - not bad for 50+ year caps! I still have the bias supply multi-cap working but I will replace that with discrete caps at some stage and leave the old cap in place (I can't source multi-caps with a common positive). I see the MB3045s have 4x22uf electrolytics in your circuit in addition to the multi-caps in the power supply. The 22uf caps could be replaced with film caps, which would lower noise... 1
couchy Posted October 18, 2023 Posted October 18, 2023 Here is a pic of one of the MB 3045 monos. 2 3
Luckiestmanalive Posted October 19, 2023 Author Posted October 19, 2023 Thanks! That is a very beautiful thing, my friend! Looking at the circuit diagram, it is far more complicated than the MQ3600/A3600, which shares some of the same tubes (8045Gs and 6240Gs). I would like to hear a pair of MB3045s in their original form with original tubes, as I would an original MQ3600. Still, I'm pretty happy with my modified version now. 1
couchy Posted October 23, 2023 Posted October 23, 2023 Hi, yesterday I did take out the old speaker connections and added banana sockets. I listened for around 4 hours. Sweet amps. I have never biased the tubes in these. I did notice one of the monos had extra caps. See image 4 vs image 5. A repair without taking the old one out I presume. They are in as good a condition as you could expect for their age. I imported these form the USA many years ago and I think over there they were just on display for many years. I do use them sparingly. 1 1
xlr8or Posted October 23, 2023 Posted October 23, 2023 7 hours ago, couchy said: Hi, yesterday I did take out the old speaker connections and added banana sockets. I listened for around 4 hours. Sweet amps. I have never biased the tubes in these. I did notice one of the monos had extra caps. See image 4 vs image 5. A repair without taking the old one out I presume. They are in as good a condition as you could expect for their age. I imported these form the USA many years ago and I think over there they were just on display for many years. I do use them sparingly. It's probably a larger cap value used in img 5 double the uF value of that in img 4 because they are wired in parallel. I would source the same ones for use on both amps. Are your MB3045 monos configured for 120v AC mains? 2
couchy Posted October 23, 2023 Posted October 23, 2023 Yes they are 120 volt but I run them around 110 as one of the transformers rattles or vibrates a bit at 120. Silent at 110v. I should have had a closer look at the cap values. Next time I open them up I will take note. I will attempt to bias them next since its never been done since I owned them.
xlr8or Posted October 23, 2023 Posted October 23, 2023 1 hour ago, couchy said: Yes they are 120 volt but I run them around 110 as one of the transformers rattles or vibrates a bit at 120. Silent at 110v. I should have had a closer look at the cap values. Next time I open them up I will take note. I will attempt to bias them next since its never been done since I owned them. My understanding is that there are 2 versions of the MB3045. One that runs at 220v AC mains configured for EU use at the time. The US version was configured for 110v AC mains use. There is also the Luxkit A3000 DIY version, which was configured for 100v AC mains for domestic distribution only in Japan around the same time. Circa 1976 and designed by Tim de Paravicini as his key highlight amp working for Luxman during this period. The 8045 and 6240 tubes are readily available on Yahoo Aleado. The 8045 was simply a GE 6550A version 2 tube reconfigured with a different pin out in the base. 2
couchy Posted October 23, 2023 Posted October 23, 2023 I was lucky enough to have a few conversations with Tim on Facebook before he passed away. May he Rest in Peace. I was so honored . 1 1
xlr8or Posted October 23, 2023 Posted October 23, 2023 4 minutes ago, couchy said: I was lucky enough to have a few conversations with Tim on Facebook before he passed away. May he Rest in Peace. I was so honored . Me too! I have a few of his units. The same Luxkit A3600 as Greg. The matching Luxkit A3300 preamp and separate Luxkit A33 power supply. A Mentmore TVA-1 power amp, EAR 864B preamp, and EAR Acute Mk1 and Mk3 CD players. 3
Luckiestmanalive Posted October 24, 2023 Author Posted October 24, 2023 (edited) Thanks, for the pics, C! Kirk is right, re the caps. In the schematic, I can see C102 is a 1uf capacitor and this matches the grey poly caps rated at 250V on the PCB on the bottom left of both IMG4&5. To the right of C102, I can see a pair of grey poly caps I imagine are C102 and C103, which are both 0.047uf caps (matching the schematic) rated at 630V. To the right of this pair there should be another poly pair that would be C105 and C106 on the schematic, which indicates they should be 0.22uF and this appears to match the two in IMG5 (I think I can make one of them out as 0.22uf at 1000V). IMG4 appears to show two pairs of 0.47uf 630V caps wired in parallel, yielding close to half that (23.5V?) at the same rated voltage. As Kirk says, replacing them in both amps with four of the same caps with same value would be a good idea. Do you know how to bias the 8045Gs? I see there are test points, TP7&8 and TP9&10, which will need two multimeters to read at the same time while you adjust the Bias pot (VR3?) and balance pot (VR2)? Edited October 24, 2023 by Luckiestmanalive
couchy Posted October 24, 2023 Posted October 24, 2023 It does have the bias instructions on the bottom of the Amp. What brand caps would you recommend? Cheers Craig.
xlr8or Posted October 24, 2023 Posted October 24, 2023 34 minutes ago, couchy said: It does have the bias instructions on the bottom of the Amp. What brand caps would you recommend? Cheers Craig. Since the 0.22uF 1000v cap represents the coupling cap in which the signal passes into just before the 8045 tube an audio grade axial type cap would be best suited for this position. @muon* should be able to advise a few suitable ones. What's your budget for a quad set? 3
muon* Posted October 24, 2023 Posted October 24, 2023 If 600v is sufficient I'd suggest Jupiter Copper/wax Foil caps for coupling (inter-stage). Compact with outer foil end indicated on the label, very nice sounding. Duelund are of course suitable also, if they fit. Miflex might be too big! https://www.hificollective.co.uk/content/jcw-060-022uf-600vdc-jupiter-copper-foil-paper-wax-capacitor As you can tell I like the natural sound characteristics of copper foil caps. Do not by-pass a quality cap like these, so if they are bypassed in the circuit remove the by-pass caps.. 3
couchy Posted October 25, 2023 Posted October 25, 2023 23 hours ago, Luckiestmanalive said: Thanks, for the pics, C! Kirk is right, re the caps. In the schematic, I can see C102 is a 1uf capacitor and this matches the grey poly caps rated at 250V on the PCB on the bottom left of both IMG4&5. To the right of C102, I can see a pair of grey poly caps I imagine are C102 and C103, which are both 0.047uf caps (matching the schematic) rated at 630V. To the right of this pair there should be another poly pair that would be C105 and C106 on the schematic, which indicates they should be 0.22uF and this appears to match the two in IMG5 (I think I can make one of them out as 0.22uf at 1000V). IMG4 appears to show two pairs of 0.47uf 630V caps wired in parallel, yielding close to half that (23.5V?) at the same rated voltage. As Kirk says, replacing them in both amps with four of the same caps with same value would be a good idea. Do you know how to bias the 8045Gs? I see there are test points, TP7&8 and TP9&10, which will need two multimeters to read at the same time while you adjust the Bias pot (VR3?) and balance pot (VR2)? Hi Greg, do you have a quality schematic of the mb3045 ?
xlr8or Posted October 25, 2023 Posted October 25, 2023 1 hour ago, couchy said: Hi Greg, do you have a quality schematic of the mb3045 ? http://ndj.g1.xrea.com/data/luxkit_b.pdf
couchy Posted October 25, 2023 Posted October 25, 2023 1 hour ago, xlr8or said: http://ndj.g1.xrea.com/data/luxkit_b.pdf Thanks a lot Kirk. Cheers. 1
Luckiestmanalive Posted October 26, 2023 Author Posted October 26, 2023 (edited) 22 hours ago, xlr8or said: http://ndj.g1.xrea.com/data/luxkit_b.pdf Wow - what a great catalogue, thanks, Kirk! It's such a shame it is in such a low resolution - I'm going to find a copy and buy it! And then I will be able to digitise it at high resolution. I can see the A3700 TCR-A7 has an ultralinear circuit with KT88s, which is how my A3600 is wired. The difference in NFB components is interesting... Also, Craig, I found this copy, sans power supply, which might be a little easier to read? Edited October 26, 2023 by Luckiestmanalive 2
Recommended Posts