Analog Sauce: The Audioromy M828-A

AKA: Tube amp part II

Audioromy M828A on the bench
Audioromy M828A on the bench

I have been fooling around with this amplifier for a month now and I have to say, it is rather fun.  There are a few hazards when purchasing Chinese HiFi (ChiFi) equipment.

The first thing to note; several places such as Ebay and Amazon list this as a single-ended class A amp. That is not true, it is a double-ended class AB amp.  I confirmed this by measuring across the two sections of the output transformer.

The second thing to note; this amp came wired with a fuse on the hot side of the AC mains and the power switch on the neutral.  Switched neutral (AKA earth, return or ground) wires are a hazard, so I rewired it, putting the switch after the fuse on the hot side after the fuse.  Another safety thing, the edge of the metal chassis was not de-burred.  I took a flat file to it and removed the burr, thus avoiding any future lacerations.

Finally; there is no manual provided with this unit.  There are a few sets of instructions on how to re-bias after tube replacement which is technically correct but not the best way to go about it. Those instructions direct the user to solder a low-value resistor from cathode to ground and then measure the voltage drop on that resistor to calculate plate current. While this is a valid way to deduce plate current, the power output tube has two tubes in one envelope and the cathodes are tied together. The plate current can be calculated for both sides, but there is nothing indicating that the two sides are balanced and one of the tubes can red plate. This was also noted in those instructions found online.

That being said, I thought I could type up a set of directions that are more suited for this amplifier.  But first, read this dire warning about working with High Voltage:

This amplifier has lethal voltages present during operation. It is possible that lethal voltages can be stored in certain components for days after the amplifier has been turned off and disconnected. By removing the protective covers, those components will be exposed and you may come in contact with them if you are not careful.

If you are planning to service this amplifier, it is vital that you have basic electronics and electrical knowledge. This includes all applicable safety procedures for working on high-voltage components.

If you do not have this knowledge, please bring this amplifier to a qualified electronics technician or repair shop for service.

I am not responsible for any injuries or damage suffered to yourself or others if you decide to undertake repairs to this equipment.

I acquired a few of these Ulyanovsk GU-29 tubes and decided to try them out.  The maximum plate dissipation for this tube is 40 watts with bulb temperature of 175°C and ambient temperature of 20°C.  I measured the bulb temperature at 142°C and the temperature in my living room ranges from 20ºC to 32ºC (68ºF to 90ºF).  I could, in theory, bias these tubes for a higher plate dissipation, if I wanted to.

Ulyanovsk GU29 tube, made in the USSR
Ulyanovsk GU29 tube, made in the USSR, circa May 1964

I asked my Russian friend what the assembly line person or factory manager might think if he or she knew that the tube made in their factory would end up being used in a home audio amplifier owned by a guy in New York.  She said, “They would have a stroke.”  Ulyanovsk had and still has a heavy military presence, thus they likely assumed that all their products would be used by the Soviet Navy or Army.

Being that this particular tube sat around in a warehouse for 55 years, it was slightly gassy.  When I first turned the amp on, there was a distinctive pink glow and a couple of small internal arcs.  It probably would have been a smart idea to light up the filaments for several hours before applying plate voltage.  Unfortunately, I had that idea after I’d already energized the amplifier.  In any case,  I increased the bias and reduced the plate current.  After a while, things settled down and I got to work re-biasing the amplifier.

To re-bias the amplifier after new tubes have been installed, some initial data needs to be gathered.  Basically, this procedure involves measuring the resistance of the plate circuit, then measuring the voltage at the output of the plate voltage supply and the voltage at each of the plate terminals on the power amp tube.  The plate voltage on this amplifier is +460 DC or so voltage above ground potential.  Obviously, this is a dangerous voltage and if you are not familiar with working on high voltages, do not attempt this procedure.  The best way to measure these test points is to use clip leads; turn the amp off, let the capacitors discharge, place the clip leads on the appropriate test points, turn the amp on, make the measurement, then turn the amp off, repeat as necessary.

After the replacement of the power tubes (V-5 and V-6), the bias for those tubes should be checked and adjusted as follows:

A. To measure plate dissipation as set by the factory, perform the following steps:

1. With the amp completely turned off and disconnected from the AC mains, remove the bottom cover. Ensure that the large power supply capacitors are discharged to the ground. With an accurate ohm meter, measure from the exposed lead on L-1 (TP-1) on the power supply board to the input to the anode resistor (R-21 or R-22 in the schematic diagram (TP-2, TP-3, TP-4, TP-5)) for each tube (four measurements total). Make a note of those measurements. For reference, my amp measured between 163 to 165 ohms.

2. Reconnect the amp to the AC mains and turn on the power (be sure to read the dire warning about high voltage above). With an accurate DVM, set to DC volts scale, carefully measure the voltage on the exposed lead of L-1 on the power supply board to the ground, and make a note of it. This is the B+ voltage for the amplifier. Carefully make another measurement between the input of the anode resistor (R-21 or R-22) and ground (four total measurements, likely to be the same), this is the plate voltage for the power tubes. Make a note of that as well.

3. Subtract the plate voltage from the B+ voltage. For my amp, this was 462 VDC 458 VDC = 4 volts. This can also be measured between TP-1 and TP2 through TP-5 See charts 1 and 2 below. This is the voltage drop. Using ohms law, calculate the plate current for each section of the amp:

Voltage drop ÷ resistance = plate current or 4.17 VDC ÷ 163.2 ohms = 0.0255 amps (25.5 ma) plate current.

Using ohms law, calculate the plate dissipation for ½ of the power tube:

Plate voltage × Plate current = Plate Dissipation or 458 V × 0.0255 amps = 11.7 watts.

Add both sides of the tube together for the total plate dissipation.

Chart 1: Left power tube, V-5

Test points Resistance Plate voltage (TP-2/3 to gnd) B+ (L1 or TP-1 to gnd) Voltage drop Plate current Dissipation
TP-1 to TP-2 163.2 ohms 458 VDC 462 VDC 4.17 VDC 25.5 ma 11.7 watts
TP-1 to TP-3 163.9 ohms 458 VDC 462 VDC 4.16 VDC 25.3 ma 11.6 watts

The total power dissipation for V-5 is 23.2 watts or 77% of the maximum for the stock FU29 tube. That is slightly above the commonly recommended safe range of 70% of the maximum, but it is tolerable.

Chart 2: Right power tube, V-6

Test points Resistance Plate voltage (TP-4/5 to gnd) B+ (L1 or TP-1 to gnd) Voltage drop Plate current Dissipation
TP-1 to TP-4 164.4 ohms 458 VDC 462 VDC 4.23 VDC 25.3 ma 11.6 watts
TP-1 to TP-5 164.8 ohms 458 VDC 462 VDC 4.14 VDC 25.1 ma 11.5 watts

The total power dissipation for V-6 is 23.1 watts or 77% of the maximum for the stock tube.

Test points for tube bias adjustment
Test points for tube bias adjustment
Pin out, 829B tube
Pinout, 829B tube

B. When replacing the power tubes, it is recommended that they be replaced in kind in pairs.

Step 1: Increase the tube bias (measured on pin 2 or 6 of the power tube) to -25 VDC and check the plate voltage drop on both tubes. Increasing the bias will reduce the plate current and thus the plate dissipation.  This will be noted as a decrease in the voltage drop.  A good starting set point would be 50-60% of the normal factory plate current (Vd ÷ Plate R) setting. The voltage drop can be measured directly by connecting the positive lead to TP-1 and then measure TP-2 to TP-5 with the negative lead. Use clip leads, placing them on the test points with the amplifier turned off.  Be extremely careful; these test points are +460 VDC above ground when the amp is energized. Read dire voltage warning above.

Step 2: Turn off amp, discharge power supply capacitors, replace tubes.

Step 3: Allow the new tubes to burn in for approximately 3-5 hours with reduced plate dissipation, make sure that the amplifier is connected to a suitable load on the speaker output terminals.

Step 4: With the DVM connected to TP-1 and TP-2, slowly bring the bias down until the plate circuit voltage drop approaches the values for the old tube.  Repeat procedure for each plate circuit (TP-1 to TP-3, TP-1 to TP-4 and TP-1 to TP-5). Recalculate plate dissipation. Be sure not to exceed plate dissipation of the tube! It is best if the tube is biased to run at about 70-75% of the maximum plate dissipation.

Step 5: With the amp fully warmed up, turn out all lights and observe the plates of both tubes for any signs of red plating.

Step 6: Carefully measure the balance between the two plate outputs of each tube by placing the DVM leads on TP2 and TP3 for V5 and TP4 and TP5 for V6. Alternatively, the test leads can be placed directly on Pu-1 and Pu-2 of the power tube under test. Between these test point pairs, the DC voltages should be zero or close to it. Note; there will be some fluctuations in the hundredths or thousandths volt ranges. Very, very carefully, adjust the bias control pots until the voltmeter reads zero or as close as you can get to zero.

Step 7: Recheck the plate dissipation for both sides of the tube, make sure that they are closely matched and not exceeding the maximum plate dissipation for the power tube in use.

I discovered several things during this process; it is very easy to red plate one side of the tube while adjusting the bias controls.  Fortunately, I noticed this right away and was able to stop the red plating quickly.  The Ulyanovsk tubes seem none the worse for wear.  As Alex Ovechkin says “Russian machine never breaks.”

Next, the schematic diagram I posted previously is not correct for this version amplifier.  There are two bias voltage controls, one for each grid.  There is no balance control, the tubes are balanced by making very careful adjustments to one or the other of the bias controls.  Updated schematic diagram:

Audioromy M-828A schematic diagram
Audioromy M-828A schematic diagram

When the amplifier is properly biased and balanced, the distortion figures should be very low, less than 0.5 to 1% THD at full power.  It makes a big difference.

The point of all this is to 1) have fun, 2) perhaps learn something about tube (or valve) circuits and 3) listen to really clean, good-sounding audio.

Tube Amp

I have dipped my toe into the world of tube (or valve) audio.  The first thing that I learned was that in general, tube amps are expensive.  It seems that the least expensive amps run about $1,000 US, and from there it seems the sky is the limit.

There are a number of less expensive Chinese versions floating around, most of the tube audio experts call them garbage.  Myself; I am not so sure.  There are also a lot of somewhat dubious claims made by the same experts about speaker cables, AC power conditioning, and so on.

I was going to build a single-ended tube amp based on the KT88 design found here:

That is a whole series of videos, eighteen in all I think, on the design and construction of a single-ended KT88 audio amp.  If you have the time, well worth it to watch.

Then I decided that I really do not have a lot of time for that and I just wanted to try a tube amp and see if there is really that big of a difference.  Thus, I purchased one of the Chinese designs based on the RCA 829B tube, which is kind of exotic looking:

FU-29 Chinese equivalent to RCA 829B dual pentode tube
FU-29 equivalent to RCA 829B dual pentode tube

That is the Chinese version FU-29, there is also a Russian радиолампа ГY-29.  The good news is that there are lots of these tubes available for not too much money.  New Old Stock (NOS) RCA 829Bs run about $25-30 each.  A Ulyanovsk GU-29 (NOS) runs about $10.00 (made in the USSR).  Somewhat rare are the 3E29 tubes, which were designed for VHF pulsed radar.  These are dual pentode tubes that can be run ether parallel (single-ended) or push-pull.  They were originally designed for VHF transmitters, but have been put into use in HF transmitters and audio amplifiers.  The USSR versions are long-life militarized versions and designed for aircraft radar; flying upside down at Mach 2 in -50 C temperatures 18,000 meters AMSL…  My Russian friend tells me I am joking.  I am joking.

Reflector factory, 6N3P-E dual triode tube
Reflector factory, 6N3P-E dual triode tube

The driver tubes and phase inverters are 6N3P-E (6N3, 6N3P, 6N3P-EV, 5670, 2C51 or 396A can also be used) which is a double triode tube, made by Reflector in Sartov, Russia.  These tubes are also militarized long-life versions.

Audioromy M828A push pull tube amp
Audioromy M828A, power transformer and output transformers

The Audioromy M-828A amplifier seemed like a good compromise between price, power, and workmanship.  I ordered the amp from Amazon and it took about a week to arrive.  The first thing I did was take it apart and look at it.  I was expecting poor workmanship and cheap components, etc.   Overall, it seems to be pretty well-made. There are two printed circuit boards; one for the power supply, the other for the front end before the two power amp tubes.  The power supply uses solid-state diodes, which some view as a compromise to a tube amp design.  There are also several power supplies on one board; 460 VDC B+ for power tubes, 220 VDC screen supply, a -25 VDC grid bias supply, 12 VDC for the audio switching relay, +6 VDC for the driver/phase inverter filaments.  I like the idea of DC filament voltage on the driver tubes.

Audioromy M828A underside
Audioromy M828A underside

This amp is configured for push-pull and rated at 30 watts per channel.  I will test all of that plus measure THD, frequency response, and so on.

There is no manual, which I find a little bit annoying.  Also, there is a lack of a schematic diagram or any instructions on biasing and balancing the tubes when they are replaced.

Being thus annoyed, I did some deep diving on the intertubes and found that some people had posted on how to re-bias and re-balance the thing after tube replacement.  There were also several modifications suggested.

  1. Replace the input potentiometer with something a little more substantial.  It does seem to be a little bit cheap and I do not like the notches in the volume adjustment.  I will do this mod.
  2. Replace the coupling caps with oil-filled units.  Not so sure about this one, but I might try it just to see if it makes a difference.
  3. Install a bias regulating circuit using an LM317 voltage regulator between the output tube cathode and ground.  This seems like a good idea.
  4. Roll (replace) the input and power tubes with better versions of US-made or Russian-made tubes.  The input tubes are 6N3P-E tubes from Reflector (Sartov, Russia) which are already pretty good tubes.  I might replace the FU-29’s with a set of GU-29’s at some point.

There appear to be several schematic diagrams with slight variations based on the changes in design over the years.  Several designs have different input and phase inverter tubes.  Some have different power supplies, still, others show no anode resistors or a cathode resistor.  This is the diagram for the amp that I own, which was produced circa 2018 or so:

Audioromy M-828A schematic diagram
Audioromy M-828A schematic diagram

Full schematic here: https://www.engineeringradio.us/blog/wp-content/uploads/2019/03/M828A_full.pdf

After all my investigations were finished, I put the amp back together and plugged it in.  I then ran my known CDs through it and it sounded a bit rough.  I was a little bit disappointed until someone said that it takes about 10 hours or so for a tube to break in.  I connected it to my speaker test load (8 ohm, 50 watt resistors) and let it run for a day.

What a difference a day makes.  The second listen to the same CD proved to be much, much better.  There is definitely some coloration from the tubes.  A side-to-side comparison between my solid-state Kenwood VR-309 amp and the Audioromy M-828A has the tube amp sounding much richer.  There is no real way to say it; it sounds full while detailed and clean all at the same time.  Playing through my homemade speakers, which are mid-range deluxe, stringed instruments sounds very detailed.  You can hear the pick hit the strings on an acoustic guitar.  You can hear the bow scrape across the strings on a cello.  It is unlike any amp that I have ever owned.

I am enjoying very much listening to Dave Mathews and Tim Reynolds Live at Luther College CD as I am typing this.

Now, I don’t know what the difference between this amp and the $10,000.00 version of the same tube amp made in Canada, other than the $9,500.00 difference in price.

A few comments about this amp and the 829B push-pull amp design.  First of all, since the screen grids are connected internally, there is no way to run this tube in ultra-linear mode.  Usually, ultra-linear mode involves taking feedback from the output (anode) or the output transformer and feeding it into the screen of the power tube.

Secondly, it is widely commented that these amps are notoriously difficult to bias and balance.  One or both sides of the output tube will red plate due to over current.  I am hopeful the LM-317 bias regulator circuit will take some of the difficulty out of this.  With an ordinary push-pull amplifier, the balancing issue is taken care of with matched tubes.  Since both tubes in this push-pull circuit are in the same envelope, getting a matched pair is not likely.  So, the tricky act of balancing the two outputs from the same tube will have to be carried out each time the tubes are replaced.  That being said, hopefully, a set of those Soviet tubes will last for a long time.

One thing that I did do is make a bunch of voltage measurements and noted them on the schematic diagram.  If there are ever any problems with this unit, having a set of base voltage measurements should go a long way toward troubleshooting and repairing it.

Finally, while the 829B is a rather exotic tube, it likely does not perform to the level of an EL86 or KT88 single-ended design.  That being said, I have no problems with purchasing this amp and I am enjoying the toob audio sound very much.

Box One

I have been remiss in updating this thing, even for Christmas and the New Year.  It has been a busy time, but also, it seems that there is nothing exciting to write about.  Continuing on writing about another transmitter installation or studio project seems redundant.

That being said, I have moved into the realm of high quality audio.  I miss that days when a good audio was the general rule, in both home audio and broadcast.  People have become used to crappy .mp3s played through crappy computer speakers or cheap ear buds.

Knowing just enough to be dangerous, I figured I should do a little bit of research before spending a lot of money foolishly.  I discovered that there are gobs and gobs of information on various forums and other places around the intertubes.  Most of it seems to be good, although one has to be careful and backup whatever is out there with science.  There are several books about DIY speaker building, amplifier construction, turntable maintenance, etc.  Picking the thing that I thought would be easiest and lead to the  biggest improvement in my own audio system, I set out to build a pair of speakers.

Most people probably don’t realize this, but there is quite a bit of work that goes into a well designed pair of speakers.  I began by thinking about what the end use will be, which eventually is a single ended tube amp based on a KT88 design.  As such, I figured the efficiency of the drivers was an important detail.  Power handling capability of the driver could be quite low, 30-50 watts or so.  Searching through several speaker manufacture’s web sites, I found a small sized, full range driver that is fairly efficient and has excellent reviews.

The Tang Band W4-1337SDF has a published sensitivity of 89dB/1 watt/1 meter.  Its frequency response is 70-20,000 Hz.  It also has a titanium speaker cone.  There have been many an article written and much ink spilled on metal cone speakers, so I did not quite know what to think of the titanium cone.  I did spend a goodly amount of time reading all of the reviews on this particular driver and decided to take the risk and buy two of them.

Next step was to calculate the proper interior volume of the speaker enclosure for a vented box.  Vented or ported speaker enclosures are generally more efficient than sealed units.  Vented boxes are a little bit more exacting to build correctly.  Again, lots of information available on line, some of it is good.  In the end, I downloaded a free software package called WinISD.

WinISD takes into account all of the Thiele/Small characteristics of the driver and generates  a basic box design.  I looked at the proposed box and decided that the internal volume was the important part, the actual shape of the box is secondary so long as it is not an exact cube.  Instead of the 2:3 ratio rectangle, I choose something different; a 1:4 rectangle.

Making cuts for speaker boxes
Making cuts for speaker boxes

Next, I began looking around at available materials.  I have plenty of wood laying around from previous projects, so I decided to make the boxes from 1×6 clear pine.  This is also contrary to conventional wisdom, as MDF is the preferred choice in speaker cabinets.  This is because natural wood has a resonate frequency, which can create problems.  As these are low power units, I figured, if it was a huge problem I could always make another cabinet out of MDF.  In the mean time, the wood, glue, paint, screws, foam insulation, tung oil finish where already in the shop.  Why buy more stuff?

I also wanted to add a tweeter (Peerless D19TD-05) to cover the high end and a simple 1 pole (or first order) cross over.

Speaker box work
Speaker box work

Thus, parts ordered, I started working on the boxes.  I decided that rabbit joints where a better choice than mitered 45 degree joints.  I used the router table to make the joints, cutouts and round the cabinet edges.  During the sanding process, I discovered that the wood boxes do indeed resonate somewhere around the 300 to 400 Hz region.  More on that later.

Speaker box glue up
Speaker box glue up

The fronts and backs are made out of 1/2 inch plywood, painted flat black.  There is a one inch rear firing port.  The box itself is larger than what is called for.  I made it thus because there where a couple of different recommendations on box volume and I wanted to add some cross bracing, which takes up space.

Speaker box, foam dampening and bracing
Speaker box, foam dampening and bracing

I thought about ways to dampen the wood box resonance and came up with a bit of rigid foam insulation, again left over from some long ago renovation project.  My idea was to take up some of that excess internal volume, but they might also work to dampen the resonance.  I cut several pieces of this material so that they fit snugly into the box.  I then used the sander to resonate the box and see what effect the foam insulation was having.  In the end, I came up one piece at the top and bottom and one approximately in the middle.  Once I was happy, these were glued in place.  This significantly dampened the resonance.  I also added quite a bit of acoustical foam inside the box.

First order crossover
First order crossover

The cross over is designed for 4000 Hz.  It consists of a 5 uF capacitor and a .31 uH inductor.  I am a minimalist at heart.  I thought about nixing the inductor altogether, but I think running both the driver and tweeter at the same time would lower the impedance too much over the high frequencies.

Completed speakers
Completed speakers

The completed project was bench tested using a software program called DATS:

Speaker impedance sweep
Speaker impedance sweep

The Tang Band driver is resonant at 60 Hz or so.  After the F3 frequency, calculated to be 101 Hz, the impedance looks good all the way out to 20 KHz.  It appears the F3 frequency is slightly higher, likely because the port is too short.

I messed around with the internal box volume by adding and taking away pieces of foam insulation.  In the end, I found that the original volume calculated by WinISD worked (and sounded) the best.

I set these up and took a listen.  Using a reference recording of Tschaikovsky (piano concert #1, B flat minor) I found these speakers sound excellent.  The stringed instruments and horns in particular sound very detailed.  The piano is open and natural.  If I close my eyes, it sounds like it is right in front of me.  Perhaps that is the wood box.  I tried them on several different types of music; jazz, rock and even Tom’s Dinner.  It may be a bit biased, however, I find these speakers to be far and above anything else I have owned in the past.  They sound great.

My only very minor gripe is the bass is not as responsive as I would like.  The low end starts around 90 Hz.  This showed up in the F3 frequency reported by WinISD.  I have a Polk Audio subwoofer that I am using (temporarily) to add the bass back into the mix.  I could also try tuning the ports a little bit to move the F3 down.  That may also require removing some if the foam from the box to increase the internal volume.

I also made a small mistake when cutting the wood for the box, as they are slightly too narrow and the driver does not fully fit onto the plywood front.  That is because I started working on this before I had the drivers in hand.  If I make another pair, I’ll make the cabinet a little bit wider.

Speaker frequency room response
Speaker frequency room response

I also ran a couple of sweeps with Room EQ Wizard.  That 300-400 Hz box resonance shows up in the sweep, but it is not noticeable when listening.  Without the subwoofer turned on, the bass does not start to pick up until about 70 Hz or so, which exactly the spec on the driver.  Funny how that works.

Speaker and subwoofer
Speaker and subwoofer frequency response

This is with the subwoofer turned on. Notice the little hum around 40 Hz, that is the hallway to the bathroom acting as a bass resonator.  Unfortunately, my listening room has some uncurable defects; I cannot get rid of the hallway to the bathroom because eventually that room comes in handy.  I need to get some acoustical material up on the wall and perhaps the ceiling.  I was thinking of a Helmholtz resonator in the wall.

Speakers mounted
Speakers mounted

They sound slightly better if they are moved off axis from the back wall.

My total cost was about $180.00, not counting the materials I already had on hand.  After listening to these for several days, I can say they stack up well against speakers that cost ten times what I paid.

Next project; the matching subwoofer.  I have some ideas…