This is a second installment (first one can be found here http://www.audiokarma.org/forums/showthread.php?t=429993) of my experience with Yaqin MC-100B amplifier, which includes measurements and notes on my attempt to tweak it for a search of better sound.
After a week of listening sessions I took Yaqin to my lab for testing and possible tweaking. First I decided to measure basic parameters like output power, THD, noise level to get sense of what you can get out of the box and set a baseline for modifications. In my measurements I used the BK Precision function generator, Rigol digital oscilloscope, EMU 0404USB audio interface with TruRTA and RMAA software packages for precise measurements. In all tests I used 8 Ohm dummy loads connected to both channels. I did all tests in Yaqin in “power amplifier” mode, using front inputs. MC-100B can be used in “integrated” amplifier mode, when it activates source switch and also increases sensitivity by reducing overall negative feedback, but I never had intention to use it that way, so I didn’t bother to try.
Initial measurement results were quite positive. Noise level was below generally below 90 dB, with slight 60 Hz hum and its harmonics visible on FFT results. But they all were below 75 dB. That corresponded well with my listening experience, where I only heard slight hum when my ears were within an inch from speaker. Yaqin does not use choke in the filter, but two 470uF capacitors in each channel (MC-100B is indeed almost completely dual mono design, with only last stage bias supply shared) do a very good job to eliminate any power supply related noise.
Next I measured bandwidth. Again there were no surprises here, and it was completely flat within audio spectrum, slow declining from 30 kHz upwards. But I found a resonant peak at around 270 kHz. This is probably due to resonance in output transformer (which is common in tube amplifiers). This shouldn’t usually be a problem unless you listen to SACD or other content with excessive ultrasound noise. Looking forward, I tried to reduce this problem with some success, I will provide details later.
Now comes most important part. When I tried to measure THD and IMD at rated power (60W), it was clear that there is something wrong. I saw all FFT screen filled with distortion products. When I connected oscilloscope to the amplifier’s output is was clear why – I saw almost square signal instead of expect 1 kHz sine. Yes it reached expected 21 Volts RMS on 8 Ohm load, but this is not was you expect from audio amplifier. So first point taken – rated power has nothing common with reality in this amplifier. I then tried to reduce volume to the point when I didn’t see obvious distortion on the oscilloscope (which usually means THD below 1%). And I found that only if output signal is below 15 Volts RMS (which corresponds to 28 Watts), I do not see much of distortion. I ran tests at that power level and got reasonable results with THD + noise at 0.7%, and IMD (using 60 + 6000 Hz, 4/1 amplitude tones) at 1.5%. But that only was at base frequency 1 kHz. When I tried to measure THD at 8 kHz, results were rather poor. Oscilloscope again showed me why – at 8 kHz, I had to reduce output signal to 12 Volts (18 Watts) to get visually correct sine wave. It all means that full range power limit of MC-100B is only around 18 Watts. I repeated measurements at 1 Watt of output power, and THD was at much better 0.07% level, which was expected.
Recalling 270 kHz peak, which I found in my bandwidth test, I decided to see if there is any possible instability. I used 1 kHz square signal from functional generator. It has rise and fall time of 20 nS. And I found what expected – there was definitely ringing with close to 300 kHz frequency, which continues for 4 – 5 cycles before settling down. There was definitely stability problem with MC-100B, which needs to be addressed. Since Yaqin uses higher feedback when in “power amplifier” mode, I repeated just this test using “integrated” mode. Ringing was still there, but reduces to 2 – 3 cycles. This confirmed that problem was with instability caused by global feedback, and when it was reduced, ringing was reduced too (this is textbook behavior).
Overall results from initial tests were in line with expectations from Chinese design and relatively cheap implementation. In out of the box form this MC-100B provides adequate performance for someone who decided to get into tubes. But there was really no reason to prefer it to any competent solid state amplifier, other than for aesthetic reason – it really looks nice with its black and chrome color palette and glass tubes glowing with orange lights. Once distraction was bright blue LED on the front panel. It would be more appropriate to use orange LED there, and I will likely do that this swap one day.
So now it was clear that Yaqin needs some work to be done on it to make it sound better. First I found schematic diagram. I could do reverse engineering on PCB myself, but someone already did it, so I used his work (I do not know what is the origin of it, so I do not know who I need to thank for it). One copy of schematic is available at http://www.ict-net.net/100B2.jpg Of cause I made sure that it matches to what I found when I opened bottom panel of Yaqin. One error just popped up – R5 resistor is indeed 510 K, not 39 K as you can see there. Another was that volume control pot is 20 KOhm, not 50 KOhm as shown on drawing. These both are minor differences, which practically do not change the sound.
When I verified that diagram is correct, I entered it into circuit simulation program (I use Spectrum Software’s Microcap 9). It has models for 12AX7 and 6SL7 tubes, but not KT88. I did extensive Internet search again and found two different SPICE models for KT88, as well as alternative models of other tubes. Microcap permits to switch models quickly, so I used them alternatively trying to match results I found in real thing. Another problem was that I didn’t have parameters of output transformer. There was no any labels on transformer (yes, I did remove black cover can), so it seems to be a custom design for this amplifier. But I found detail information on several output transformer for KT88 tubes available from different vendors and tried to use closest I term of winding resistance (I do not have LCR meter on hand to measure real thing, maybe someone will do it and publish results, which will be helpful for other tweakers).
Now having the results from initial simulation run, I started with measuring actual amplifier. First thing I noticed is unexpectedly B+ level. From diagram you can see that secondary winding of power transformer should deliver 360V, which corresponds to 509 Volts expected on B+ bus. But I saw only 410 Volts. I checked by line power and it was at 117 Volts, which is close to nominal 120. Moreover, filament heater supply was close to 6.3Volts. Which means that power transformer works as specified (I actually calculated that it was likely designed with 115 Volts source in mind). I recall reading messages on some forums about Yaqin amplifiers supplying 6.9 Volts to filaments, but this is not my case. So it is likely that Yaqin redesigned their power transformers my increasing primary windings. Actually if I potentially increase source voltage to get 6.9 Volts at filament, it will have specified 360 Volts on main secondary.
I did find several application notes for KT88 tube, and they always specified more than 500 Volts of the plate to get 60 – 80 Watts of power. In my case I only had 400, which explained power limit at half of power KT88 is able to deliver. There is no real workaround for Yaqin design, other than replacing power transformers. But this is expensive proposition, as they are sold for $120-160 each (and you need two for dual mono design like MC-100B). So the only I could do with tweaking is to make sure that I squeezed last drop of power from Yaqin and make sure that all other stages are as linear as possible.
After I adjusted power supply parameters in Microcap model equal to what I measured, I got a reasonable match between DC simulation results and actual measurements. Difference was less than 10%, which I consider accurate enough.
Now it was time to do some optimization. Yaqin uses SRPP circuit as first stage. Usually SRPP topology is used for a relatively low impedance load, but here it has 1 MOhm load from Schottky phase splitter. There are two variables you can play with here: R3 and R4 resistors. From commonly known SRPP stage calculations for 1M load, you would expect R3 and R4 to be 2.4 KOhm. But simulation with goal to get highest output with lowest THD resulted in different values: R4 at 1K and R3 at 3K. This change will significantly reduce bias at grid of Q2 (MC-100B uses direct connections though first three stages, probably trying to avoid using capacitors in the signal path), so to keep idle current of Schottky stage, it was reasonable to reduce R8 to 43K. Another possible improvement lies in phase splitter plate load resistors. Yaqin decided to use the same 47 KOhm resistors in both sections of it. But it is recommended to have R7 higher (or R6 lower) to balance the circuit. Simulation results recommended either 43K/47K or 47K/51K pair. It was hard to simulate final stage without exact specification output transformer, but using reasonable guess, I got results close to expected for maximum output power in 30 – 40 Watts range.
Now it was time to implement changes in real amplifier and see what I can get. This will be a topic of my third article.
Part 3: http://www.audiokarma.org/forums/showthread.php?t=430113
After a week of listening sessions I took Yaqin to my lab for testing and possible tweaking. First I decided to measure basic parameters like output power, THD, noise level to get sense of what you can get out of the box and set a baseline for modifications. In my measurements I used the BK Precision function generator, Rigol digital oscilloscope, EMU 0404USB audio interface with TruRTA and RMAA software packages for precise measurements. In all tests I used 8 Ohm dummy loads connected to both channels. I did all tests in Yaqin in “power amplifier” mode, using front inputs. MC-100B can be used in “integrated” amplifier mode, when it activates source switch and also increases sensitivity by reducing overall negative feedback, but I never had intention to use it that way, so I didn’t bother to try.
Initial measurement results were quite positive. Noise level was below generally below 90 dB, with slight 60 Hz hum and its harmonics visible on FFT results. But they all were below 75 dB. That corresponded well with my listening experience, where I only heard slight hum when my ears were within an inch from speaker. Yaqin does not use choke in the filter, but two 470uF capacitors in each channel (MC-100B is indeed almost completely dual mono design, with only last stage bias supply shared) do a very good job to eliminate any power supply related noise.
Next I measured bandwidth. Again there were no surprises here, and it was completely flat within audio spectrum, slow declining from 30 kHz upwards. But I found a resonant peak at around 270 kHz. This is probably due to resonance in output transformer (which is common in tube amplifiers). This shouldn’t usually be a problem unless you listen to SACD or other content with excessive ultrasound noise. Looking forward, I tried to reduce this problem with some success, I will provide details later.
Now comes most important part. When I tried to measure THD and IMD at rated power (60W), it was clear that there is something wrong. I saw all FFT screen filled with distortion products. When I connected oscilloscope to the amplifier’s output is was clear why – I saw almost square signal instead of expect 1 kHz sine. Yes it reached expected 21 Volts RMS on 8 Ohm load, but this is not was you expect from audio amplifier. So first point taken – rated power has nothing common with reality in this amplifier. I then tried to reduce volume to the point when I didn’t see obvious distortion on the oscilloscope (which usually means THD below 1%). And I found that only if output signal is below 15 Volts RMS (which corresponds to 28 Watts), I do not see much of distortion. I ran tests at that power level and got reasonable results with THD + noise at 0.7%, and IMD (using 60 + 6000 Hz, 4/1 amplitude tones) at 1.5%. But that only was at base frequency 1 kHz. When I tried to measure THD at 8 kHz, results were rather poor. Oscilloscope again showed me why – at 8 kHz, I had to reduce output signal to 12 Volts (18 Watts) to get visually correct sine wave. It all means that full range power limit of MC-100B is only around 18 Watts. I repeated measurements at 1 Watt of output power, and THD was at much better 0.07% level, which was expected.
Recalling 270 kHz peak, which I found in my bandwidth test, I decided to see if there is any possible instability. I used 1 kHz square signal from functional generator. It has rise and fall time of 20 nS. And I found what expected – there was definitely ringing with close to 300 kHz frequency, which continues for 4 – 5 cycles before settling down. There was definitely stability problem with MC-100B, which needs to be addressed. Since Yaqin uses higher feedback when in “power amplifier” mode, I repeated just this test using “integrated” mode. Ringing was still there, but reduces to 2 – 3 cycles. This confirmed that problem was with instability caused by global feedback, and when it was reduced, ringing was reduced too (this is textbook behavior).
Overall results from initial tests were in line with expectations from Chinese design and relatively cheap implementation. In out of the box form this MC-100B provides adequate performance for someone who decided to get into tubes. But there was really no reason to prefer it to any competent solid state amplifier, other than for aesthetic reason – it really looks nice with its black and chrome color palette and glass tubes glowing with orange lights. Once distraction was bright blue LED on the front panel. It would be more appropriate to use orange LED there, and I will likely do that this swap one day.
So now it was clear that Yaqin needs some work to be done on it to make it sound better. First I found schematic diagram. I could do reverse engineering on PCB myself, but someone already did it, so I used his work (I do not know what is the origin of it, so I do not know who I need to thank for it). One copy of schematic is available at http://www.ict-net.net/100B2.jpg Of cause I made sure that it matches to what I found when I opened bottom panel of Yaqin. One error just popped up – R5 resistor is indeed 510 K, not 39 K as you can see there. Another was that volume control pot is 20 KOhm, not 50 KOhm as shown on drawing. These both are minor differences, which practically do not change the sound.
When I verified that diagram is correct, I entered it into circuit simulation program (I use Spectrum Software’s Microcap 9). It has models for 12AX7 and 6SL7 tubes, but not KT88. I did extensive Internet search again and found two different SPICE models for KT88, as well as alternative models of other tubes. Microcap permits to switch models quickly, so I used them alternatively trying to match results I found in real thing. Another problem was that I didn’t have parameters of output transformer. There was no any labels on transformer (yes, I did remove black cover can), so it seems to be a custom design for this amplifier. But I found detail information on several output transformer for KT88 tubes available from different vendors and tried to use closest I term of winding resistance (I do not have LCR meter on hand to measure real thing, maybe someone will do it and publish results, which will be helpful for other tweakers).
Now having the results from initial simulation run, I started with measuring actual amplifier. First thing I noticed is unexpectedly B+ level. From diagram you can see that secondary winding of power transformer should deliver 360V, which corresponds to 509 Volts expected on B+ bus. But I saw only 410 Volts. I checked by line power and it was at 117 Volts, which is close to nominal 120. Moreover, filament heater supply was close to 6.3Volts. Which means that power transformer works as specified (I actually calculated that it was likely designed with 115 Volts source in mind). I recall reading messages on some forums about Yaqin amplifiers supplying 6.9 Volts to filaments, but this is not my case. So it is likely that Yaqin redesigned their power transformers my increasing primary windings. Actually if I potentially increase source voltage to get 6.9 Volts at filament, it will have specified 360 Volts on main secondary.
I did find several application notes for KT88 tube, and they always specified more than 500 Volts of the plate to get 60 – 80 Watts of power. In my case I only had 400, which explained power limit at half of power KT88 is able to deliver. There is no real workaround for Yaqin design, other than replacing power transformers. But this is expensive proposition, as they are sold for $120-160 each (and you need two for dual mono design like MC-100B). So the only I could do with tweaking is to make sure that I squeezed last drop of power from Yaqin and make sure that all other stages are as linear as possible.
After I adjusted power supply parameters in Microcap model equal to what I measured, I got a reasonable match between DC simulation results and actual measurements. Difference was less than 10%, which I consider accurate enough.
Now it was time to do some optimization. Yaqin uses SRPP circuit as first stage. Usually SRPP topology is used for a relatively low impedance load, but here it has 1 MOhm load from Schottky phase splitter. There are two variables you can play with here: R3 and R4 resistors. From commonly known SRPP stage calculations for 1M load, you would expect R3 and R4 to be 2.4 KOhm. But simulation with goal to get highest output with lowest THD resulted in different values: R4 at 1K and R3 at 3K. This change will significantly reduce bias at grid of Q2 (MC-100B uses direct connections though first three stages, probably trying to avoid using capacitors in the signal path), so to keep idle current of Schottky stage, it was reasonable to reduce R8 to 43K. Another possible improvement lies in phase splitter plate load resistors. Yaqin decided to use the same 47 KOhm resistors in both sections of it. But it is recommended to have R7 higher (or R6 lower) to balance the circuit. Simulation results recommended either 43K/47K or 47K/51K pair. It was hard to simulate final stage without exact specification output transformer, but using reasonable guess, I got results close to expected for maximum output power in 30 – 40 Watts range.
Now it was time to implement changes in real amplifier and see what I can get. This will be a topic of my third article.
Part 3: http://www.audiokarma.org/forums/showthread.php?t=430113
Last edited:
