Magnavox A531 Germanium Console Amp Reborn as a Vintage Stand-Alone Hi-Fi Component
Germanium transistor amplifiers are fascinating. They were proudly the first solid state designs, yet are an almost forgotten footnote lost between the chapters of vacuum tubes and silicon transistors. But germanium is worth exploring- in the same way that previously outmoded vacuum tube amplifiers are worth exploring, even though they too, had obvious design short-comings compared to the venerable silicon transistor. The germanium sound has been compared frequently with tube sound, and most of this sonic "coloration" comes from the same ingredient as tube amps- audio transformers in the signal path. Transformers provide for a fat warm bass, well defined, punchy but non-fatiguing midrange, and a delicate, silky high end. This treatment can be very beneficial when paired with modern digital sources, which are often described as harsh, technical, or sterile. I find the germanium sound to be very pleasant, and with the right music intoxicating. Those brave enough to explore germanium I think will end up giving it a welcome place among the pallet of flavors we chose from in our day to day listening.
This thread covers the restoration of a 1962 Magnavox germanium console/record player amplifier chassis to turn it into an attractive stand-alone, dare I say, "high fidelity" amplifier component. This, the first part, will cover cosmetic and chassis restoration, and the second part soon to come will cover the electrical restoration, transistor testing, and circuit details.
I have been tinkering lately with the 12V germanium amplifiers found in affordable 8-track car players from the late 60's and early 70's. They can be found almost anywhere, from garage sales to Ebay for under $30 including shipping. These really are hidden gems- because they include all of the impossible to find parts (the interstage transformers, thermistors, and matching sets of germanium pre and power transistors) already in one functional package ready for your study and redesign. Typically being of Japanese design, they are generally high quality amplifiers in a very small package. However, they are pretty low power- With only 12-14V to work with, a class AB can only generate about 4-5 Watts per channel even into 4 Ohm speakers. This is perfect for near-field listening on a desk top or bookshelf, but is not for general full-room listening. A fun detailed example of one of these that I reworked into a small stand-alone test amplifier is detailed in my "Germanium Bake King" project.
I needed something with a more power, in the 15 to 30 Watt per channel range to drive my larger vintage 3-way speakers for full room listening, and a console amplifier chassis seemed just the ticket. I had previously rebuilt a Magnavox 9300 series 6BQ5/EL84 push-pull tube amplifier chassis to be stand-alone, and thanks to the amazing Dave Gillespie redesign of this chassis posted on AudioKarma it became quite the audiophile contender and my favorite amplifier. Reworking a Magnavox germanium amp would be a perfect compliment in my collection.
When Magnavox first released their new solid state "Astro Sonic" line of consoles in the early 60's, they used germanium power transistor output stages, driven through interstage signal transformers. As these consoles age and are discarded, they provide a ready source for workable examples that can either be converted directly to stand-alone, or harvested for otherwise impossible to find parts to be reused. Many of the later models incorporated the power amplifiers into the radio chassis so they are bulky and not readily converted to stand-alone, but are a good source of parts. There are only a few designs that used a stand-alone amplifier chassis separate from the radio, primarily from the early 60's, such as the one I chose for this project.
I found this 1962 germanium amp chassis from a Magnavox record player for sale on ebay. The seller offered it for $25 plus $25 shipping, and I just could not resist. It looked pretty nasty, but would be a perfect fit if I could restore it. The seller said it did work, so I knew the bones were good, so it probably just needed a full re-cap and some minor upgrades. This is a Magnavox chassis model A531-01-00, which used the same chassis box as it's bigger brother, the A551-01-00. (In Magnavox chassis numbers, A is for Amp, A531 is the model, -01 is the version, and the -00 is revision) The more powerful A551 was used in larger full size Astro Sonic consoles with both radio and phono and was rated at 25 Watts per channel. It had a larger power transformer and used 47P1B transistors, and larger filter and output capacitors.
An example of the bigger A551:
My A531 chassis, with the smaller power transformer and the 38P1C transistors, was used in smaller stereo record player only consoles of unspecified output power. Given the supply voltage is -38V instead of the A551's -44V, I estimate about 18 Watts per channel or so. Still plenty big enough for what I need, and a good companion to the 15 Watt per channel 9300 series 6BQ5 tube amp. The interstage transformers were stamped 1386235, the 138 meaning Stancore manufactured, so excellent transformers, and the date code of work week 35 of 1962. The power transformer was also Stancore and had a similar date code work week 31. The transistors looked pretty roached- badly darkened and corroded, but they did work. The chassis still had the original molex type connector to the radio/phonograph, and the spade type connectors for speakers and pilot lamp.
To begin the cosmetic restoration, I first documented everything with photographs and notes, especially under the chassis. I disconnected and removed the power transformer discarding the really ugly looking screws and sheet metal nuts, then carefully removed and stored the transistors and heat sink to clean everything up. The only thermal paste present under the transistors was a thin oily residue, and dirt had made it's way under the transistors and the mica slips were in bad condition. This is an important area to attend to in older equipment with exposed transistors, which can benefit greatly from new mica slips and modern thermal paste.
What I feared was heavy corrosion on the nickel plated chassis turned out to be only a light fuzzy haze of corrosion, mixed with baked on dust. I tried cleaning first- alcohol, mineral spirits, and even acetone hardly made a dent. I found the fuzz was most easily removed by scraping- with the sharp edge of several wooden clothes pins that did not scratch the nickel. It was slow going, inch at a time, swapping out clothes pins as the edge dulled and started to grind into the nickel. That was followed by a vigorous rub-down with a terrycloth rag soaked in white vinegar to dissolve what I could, dried, then cleaned thoroughly with a water damp rag to remove the acid from the vinegar. While the nickel took on a nice reflective shine, it was still hazy-cloudy where corroded. At this point I could have started using metal polish like Brasso to grind down through the haze, but I don't like the shabby-schick look of over-polished nickel- it never looks right. I am reminded that with most of my projects, it's about what's under the hood and preserving it's original vintage character, not how fancy it's decorated that I really care about. It's vintage, and a bare knuckle nuts and bolts chassis, so let it be what it is in humble and dignified authenticity.
The remaining light surface rust spots on the nickel plate, rivets, and tops of the transformers came off pretty easily using a fine brass bristle brush from Harbor Freight. The brass did not scratch the steel or nickel, but removed the vinegar-softened rust down to a really nice looking patina with no damage. Be careful around the cloth covered wires and the paper/plastic wrapping on the audio transformer windings, as they damage easily even with the brush. I clipped the wires and disconnected the molex connectors, drilled out the rivets holding the the phenolic spade connector strips for the speaker and pilot connections.
I see a lot of different ways of dealing with extraneous holes in the chassis, some not very pretty. I find the most attractive treatment is to tack sheet metal patches from the inside. I keep a stash of vintage sheet metal pieces from the junk chassis I cannibalize, to cut patches for the holes. Vintage cadmium and nickel plated is excellent for soldering and is the easiest, but galvanized will work in a pinch. Fortunately I have an old monster 250W soldering gun from the thrift store that works wonders for soldering terminal strips and patches directly to the steel chassis, but a regular 140W Weller will work if you are patient. First clean the underside surface around the chassis hole with a scrap of sand paper or green scrubbie (don't use steel wool unless you like random shorts!), then add a few drops or dabs of soldering rosin. Once the plates are tacked in place, you can drill new holes for a power switch or input connectors, etc. I prefer to have the new device sit within the confines of the hole- Overlapping over the outside looks tacky to me.
The original standard Magnavox RCA input connectors are very close together, preventing you from using almost all RCA patch cables. I have found that the easiest and most attractive way to remedy this is to remove them, put in a patch plate, then drill for two new individual RCA connectors that are spread apart as far as possible while still fitting inside the original opening. This will allow enough space to use all but the fattest RCA cables.
I really liked how the appearance worked out on my 9300 tube amplifier, which kept very clean and understated, and put on an oak plinth with brushed nickel drawer pull handles, so I did the same thing here. The oak came from the center of a $3 used cabinet door from our local Habitat for Humanity which I cut down, sanded, and treated with the Natural shade of Watco Danish Oil. I really love that stuff! The drawer pulls were the most plain jane brushed nickel from Home Depot. I cut a sheet metal shield plate to sit between the chassis and plinth from some HVAC scrap also from habitat. The shield is held down flat on the oak with two small screws near the center to keep it from bowing up and touching something. For speaker outputs I chose the same type of small profile panel mount 4mm banana jacks as I did with the 9300. I really love the vintage understated look, and prefer ease of banana plugs for most of my speaker cables.
Stay tuned for part 2 coming soon: Electrical restoration, transistor testing, and circuit details
Germanium transistor amplifiers are fascinating. They were proudly the first solid state designs, yet are an almost forgotten footnote lost between the chapters of vacuum tubes and silicon transistors. But germanium is worth exploring- in the same way that previously outmoded vacuum tube amplifiers are worth exploring, even though they too, had obvious design short-comings compared to the venerable silicon transistor. The germanium sound has been compared frequently with tube sound, and most of this sonic "coloration" comes from the same ingredient as tube amps- audio transformers in the signal path. Transformers provide for a fat warm bass, well defined, punchy but non-fatiguing midrange, and a delicate, silky high end. This treatment can be very beneficial when paired with modern digital sources, which are often described as harsh, technical, or sterile. I find the germanium sound to be very pleasant, and with the right music intoxicating. Those brave enough to explore germanium I think will end up giving it a welcome place among the pallet of flavors we chose from in our day to day listening.
This thread covers the restoration of a 1962 Magnavox germanium console/record player amplifier chassis to turn it into an attractive stand-alone, dare I say, "high fidelity" amplifier component. This, the first part, will cover cosmetic and chassis restoration, and the second part soon to come will cover the electrical restoration, transistor testing, and circuit details.
I have been tinkering lately with the 12V germanium amplifiers found in affordable 8-track car players from the late 60's and early 70's. They can be found almost anywhere, from garage sales to Ebay for under $30 including shipping. These really are hidden gems- because they include all of the impossible to find parts (the interstage transformers, thermistors, and matching sets of germanium pre and power transistors) already in one functional package ready for your study and redesign. Typically being of Japanese design, they are generally high quality amplifiers in a very small package. However, they are pretty low power- With only 12-14V to work with, a class AB can only generate about 4-5 Watts per channel even into 4 Ohm speakers. This is perfect for near-field listening on a desk top or bookshelf, but is not for general full-room listening. A fun detailed example of one of these that I reworked into a small stand-alone test amplifier is detailed in my "Germanium Bake King" project.
I needed something with a more power, in the 15 to 30 Watt per channel range to drive my larger vintage 3-way speakers for full room listening, and a console amplifier chassis seemed just the ticket. I had previously rebuilt a Magnavox 9300 series 6BQ5/EL84 push-pull tube amplifier chassis to be stand-alone, and thanks to the amazing Dave Gillespie redesign of this chassis posted on AudioKarma it became quite the audiophile contender and my favorite amplifier. Reworking a Magnavox germanium amp would be a perfect compliment in my collection.
When Magnavox first released their new solid state "Astro Sonic" line of consoles in the early 60's, they used germanium power transistor output stages, driven through interstage signal transformers. As these consoles age and are discarded, they provide a ready source for workable examples that can either be converted directly to stand-alone, or harvested for otherwise impossible to find parts to be reused. Many of the later models incorporated the power amplifiers into the radio chassis so they are bulky and not readily converted to stand-alone, but are a good source of parts. There are only a few designs that used a stand-alone amplifier chassis separate from the radio, primarily from the early 60's, such as the one I chose for this project.
I found this 1962 germanium amp chassis from a Magnavox record player for sale on ebay. The seller offered it for $25 plus $25 shipping, and I just could not resist. It looked pretty nasty, but would be a perfect fit if I could restore it. The seller said it did work, so I knew the bones were good, so it probably just needed a full re-cap and some minor upgrades. This is a Magnavox chassis model A531-01-00, which used the same chassis box as it's bigger brother, the A551-01-00. (In Magnavox chassis numbers, A is for Amp, A531 is the model, -01 is the version, and the -00 is revision) The more powerful A551 was used in larger full size Astro Sonic consoles with both radio and phono and was rated at 25 Watts per channel. It had a larger power transformer and used 47P1B transistors, and larger filter and output capacitors.
An example of the bigger A551:
My A531 chassis, with the smaller power transformer and the 38P1C transistors, was used in smaller stereo record player only consoles of unspecified output power. Given the supply voltage is -38V instead of the A551's -44V, I estimate about 18 Watts per channel or so. Still plenty big enough for what I need, and a good companion to the 15 Watt per channel 9300 series 6BQ5 tube amp. The interstage transformers were stamped 1386235, the 138 meaning Stancore manufactured, so excellent transformers, and the date code of work week 35 of 1962. The power transformer was also Stancore and had a similar date code work week 31. The transistors looked pretty roached- badly darkened and corroded, but they did work. The chassis still had the original molex type connector to the radio/phonograph, and the spade type connectors for speakers and pilot lamp.
To begin the cosmetic restoration, I first documented everything with photographs and notes, especially under the chassis. I disconnected and removed the power transformer discarding the really ugly looking screws and sheet metal nuts, then carefully removed and stored the transistors and heat sink to clean everything up. The only thermal paste present under the transistors was a thin oily residue, and dirt had made it's way under the transistors and the mica slips were in bad condition. This is an important area to attend to in older equipment with exposed transistors, which can benefit greatly from new mica slips and modern thermal paste.
What I feared was heavy corrosion on the nickel plated chassis turned out to be only a light fuzzy haze of corrosion, mixed with baked on dust. I tried cleaning first- alcohol, mineral spirits, and even acetone hardly made a dent. I found the fuzz was most easily removed by scraping- with the sharp edge of several wooden clothes pins that did not scratch the nickel. It was slow going, inch at a time, swapping out clothes pins as the edge dulled and started to grind into the nickel. That was followed by a vigorous rub-down with a terrycloth rag soaked in white vinegar to dissolve what I could, dried, then cleaned thoroughly with a water damp rag to remove the acid from the vinegar. While the nickel took on a nice reflective shine, it was still hazy-cloudy where corroded. At this point I could have started using metal polish like Brasso to grind down through the haze, but I don't like the shabby-schick look of over-polished nickel- it never looks right. I am reminded that with most of my projects, it's about what's under the hood and preserving it's original vintage character, not how fancy it's decorated that I really care about. It's vintage, and a bare knuckle nuts and bolts chassis, so let it be what it is in humble and dignified authenticity.
The remaining light surface rust spots on the nickel plate, rivets, and tops of the transformers came off pretty easily using a fine brass bristle brush from Harbor Freight. The brass did not scratch the steel or nickel, but removed the vinegar-softened rust down to a really nice looking patina with no damage. Be careful around the cloth covered wires and the paper/plastic wrapping on the audio transformer windings, as they damage easily even with the brush. I clipped the wires and disconnected the molex connectors, drilled out the rivets holding the the phenolic spade connector strips for the speaker and pilot connections.
I see a lot of different ways of dealing with extraneous holes in the chassis, some not very pretty. I find the most attractive treatment is to tack sheet metal patches from the inside. I keep a stash of vintage sheet metal pieces from the junk chassis I cannibalize, to cut patches for the holes. Vintage cadmium and nickel plated is excellent for soldering and is the easiest, but galvanized will work in a pinch. Fortunately I have an old monster 250W soldering gun from the thrift store that works wonders for soldering terminal strips and patches directly to the steel chassis, but a regular 140W Weller will work if you are patient. First clean the underside surface around the chassis hole with a scrap of sand paper or green scrubbie (don't use steel wool unless you like random shorts!), then add a few drops or dabs of soldering rosin. Once the plates are tacked in place, you can drill new holes for a power switch or input connectors, etc. I prefer to have the new device sit within the confines of the hole- Overlapping over the outside looks tacky to me.
The original standard Magnavox RCA input connectors are very close together, preventing you from using almost all RCA patch cables. I have found that the easiest and most attractive way to remedy this is to remove them, put in a patch plate, then drill for two new individual RCA connectors that are spread apart as far as possible while still fitting inside the original opening. This will allow enough space to use all but the fattest RCA cables.
I really liked how the appearance worked out on my 9300 tube amplifier, which kept very clean and understated, and put on an oak plinth with brushed nickel drawer pull handles, so I did the same thing here. The oak came from the center of a $3 used cabinet door from our local Habitat for Humanity which I cut down, sanded, and treated with the Natural shade of Watco Danish Oil. I really love that stuff! The drawer pulls were the most plain jane brushed nickel from Home Depot. I cut a sheet metal shield plate to sit between the chassis and plinth from some HVAC scrap also from habitat. The shield is held down flat on the oak with two small screws near the center to keep it from bowing up and touching something. For speaker outputs I chose the same type of small profile panel mount 4mm banana jacks as I did with the 9300. I really love the vintage understated look, and prefer ease of banana plugs for most of my speaker cables.
Stay tuned for part 2 coming soon: Electrical restoration, transistor testing, and circuit details
What a great project. Looks great. Your hard work really paid off. How does it sound?
Thank you- I appreciate it. It was a tremendous amount of fun.
Well, it sounds, frankly, amazing. I don't want to steal anything away from part 2, but I can't help but gush a little- The flavor is as different from silicon as my tube amp is. I can listen to this all day, every day, and never get fatigued. When I mention "intoxicating", this is the amp that does that. Heavy metal sounds brand new again, after listening to it most of my life, and is about half of my catalog- primarily Tool, Black Sabbath, and the like. The guitars sound mesmerizing, having both a delicate detailed fuzz that is not at all harsh, and an almost 3-D holographic effect with deep sound stage that always startles me. Portishead and Gorillaz is a guilty pleasure, with that powerful deep warm sub-bass complimenting such delicate highs. A lot more of the catalog is classical- mostly strings, chamber and symphony. The highs and powerful bass are again intoxicating, picking out details in the strings and the room that I miss on most other systems. I'm seriously thinking about adding a headphone jack to this, but I really hate the idea of hacking on the chassis any more.
Thank you again, for checking it out! -W
Magnavox A531 Germanium Console Amp Reborn as a Stand-Alone Hi-Fi Component
Part 2: Circuit Walk Through
Continuation from Part 1, which covered chassis and cosmetic restoration.
I would like to say, before the following discussion, that I am NOT an amplifier expert. WHEN I am wrong, or you disagree with my interpretation, PLEASE chime in and educate me. We are all here to learn, so please help. I also apologize if I tend to over-explain- every person has a different level of knowledge, and I want to make the topic as approachable as possible. With that said . . .
The very first step in any vintage chassis restoration is to find service literature. The very experienced will be able to figure out the important stuff by eye alone, but for most this approach usually equates to pot-shotting, so if this is a serious project a schematic really is a must. I could only find one (free, I'm cheap) schematic online for this model A531, a bad scan from the Sams photofact on this chassis. I ran it through a free AI image enhancer, so it is a little more clear and it's attached. (A few alternate component values for the A551 and sub-versions are indicated in parenthesis.) I also found a nice schematic for the bigger very similar A551 chassis from Beitman's "Most-Often-Needed" schematics 1965 radio volume, and I like how that one is drawn better (also attached). Finding both lacking for use in discussion, I hand drew the schematic below to clarify as much as possible. The values shown on that are not all original- capacitors have been up-sized and some resistors changed.
This germanium amplifier uses all PNP transistors, so the current flow through the transistors feels backwards to me, from the emitter to the collector. The schematic could have been drawn conventionally with the emitters at the top. However, to make the circuit feel more familiar to those used to NPN devices (with the supply voltage at the top and the emitters at the bottom, this schematic places the negative supply rail at the top, so the emitters and GND are in the familiar place. Circuit structure will then be more recognizable, and tracing can proceed with little regard to the fact that they are PNP devices.
In studying the schematic, the design is fairly simple and easily broken down. The first stage (Q1) is an unusual looking emitter follower that has unity gain just for buffering the very high impedance input signal to present a good low impedance copy to the driver stage Q2, and to an optional line level output that is shown in the attached A551 Beitman's schematic. The Q1 emitter follower circuit looks unusual because of the additional 10K and 100uF capacitor, there to greatly increase the input impedance into Q1, and the 22K and .001uF cap in the collector circuit are there to limit the voltage dropped across the the low voltage germanium transistor. I had to reach out to the diyAudio community for help understanding this one, and you can see the discussion there if you are interested. The Q1 emitter also provides the bias voltage for Q2 of around -7V.
Q2 is a common emitter voltage amplifier providing the voltage gain of the amplifier. It has the global negative feedback signal from the speaker output coming back to it's the emitter resistance, and the collector drives the primary of the interstage transformer, coupling this stage to the power output stage. More on this later.
Most are familiar with class AB output stages using complementary drivers (NPN on bottom and PNP on top) which allows both drivers to be driven with the same phase signal. This design uses a totem pole output stage, with two PNP germanium power transistors (Q3,Q4) in a stack, requiring the two input signals be of opposite phase. The interstage transformer provides for an impedance match between the Q2 high gain driver and the heavy output transistor load, as well as delivering for free opposite phase inputs to the bases of the totem drivers. Today the use of a transformer seems quite odd, but keep in mind in the mid-1960's, capacitors and transistors were more expensive, the germanium transistors had very little current gain, and transformers were plentiful and relatively cheap, so this was a pretty valid strategy.
The bases of the output transistors are biased with a resistor divider ladder, with the transformer secondaries picking off DC voltages from appropriate rungs of the ladder to deliver both DC bias and AC signal to each base. The 3.3Ω bias resistor from base to emitter of each Q3 and Q4 sets the actual device bias and quiescent output current, in this case about 50-60mA. In other germanium transistor amplifier designs (particularly automotive amps like my Germanium Bake King ) I have seen the use of negative tempco (NTC) thermistors in parallel these bias resistor to prevent thermal runaway of the highly temperature sensitive germanium outputs. These thermistors decrease resistance as temperature increases, turning down the bias voltage. In this amplifier the liberal heat-sinking provided by the chassis, intended low power output of this design into fixed speakers, and the indoor use (as opposed to a hot car in summer) negates the need for this.
The 330Ω and 270Ω divider resistors position the DC quiescent output voltage to be supply/2, or about -18V, so the speaker is capacitively coupled. The 0.47Ω emitter resistors are of course the standard emitter degeneration resistor, used to stabilize gain and bias of the output device. The global negative feedback signal (the 1KΩ in parallel the 0.005µF cap) is picked off post output capacitor, delivering only AC signal to the feedback insertion point at the emitter of Q2.
Here is a schematic of one channel from the Sam's photofact:
In this Sam's or Beitman's schematic, there is shown a 200Ω balance pot with each side connecting to the negative feedback insertion point near the emitter of Q2 for each channel, with the center wiper connected to ground. This is an efficient but counter intuitive balance approach, something I have seen in multiple Magnavox designs. The balance control works to both change the gain of the channel, and vary the amount of negative feedback each channel is receiving. The resistors at the emitter of Q2 determine the gain and DC bias current of the stage- R20 (820Ω) sets the DC bias current (~8mA) but is bypassed by C12 (100µF) so it is shorted for AC. Therefore the AC gain of Q2 is set by R21 (33Ω) in parallel with half (100Ω) of the balance pot if set at 50%. The smaller this parallel resistance, the higher the gain of Q2. The negative feedback signal is delivered to the top of R21, which is subtracted from the input signal at the base of Q2, working to reduce the overall gain of the amplifier. (Less negative feedback, more gain). Therefore, moving the balance pot wiper towards a channel puts a smaller and smaller resistor in parallel the 33Ω emitter resistor increasing the gain of Q2, while simultaneously reducing the amount of negative feedback signal that is subtracted, also acting to increase the overall gain. With the balance pot turned all the way to that channel, the gain of Q2 is at a maximum, and the channel is running wide open "open loop" with no negative feedback, so the sound of the channel changes significantly becoming "bloated and boomy" with an audible level of distortion.
In general, a balance pot has no place being on a power amplifier chassis, especially one that changes the sound and increases distortion. Magnavox undoubtedly put this here (out of the reach of the consumer) because the speakers in this console were fixed, built in, and if there were any perceptible imbalance due to loose component tolerances it was tweaked here once and shipped. A far cheaper solution for them than using tighter component values. Tone controls, and even volume for the most part, should be in the pre-amp, not the power amp. One of the first changes I made to the chassis was to remove the 200Ω balance pot, and replaced the 33Ω resistor with a 25Ω 1% tolerance metal film resistor fixing the balance at 50% and matching this critical component between the two channels. I do however like a volume control on my power amplifiers, for those times I am running a line level source directly into it. When I use a pre-amp the volume control will be at maximum, and generally have no impact on the circuit anyway. For this design, I chose a 10K audio (logarithmic) taper. Any modern solid state source or pre-amp will typically have a low output impedance (1KΩ or less) so I find 10K volume pot to be a good trade-off. I find using any higher (people seem to prefer 100KΩ or much higher) to be induced noise prone and make more wiper noise.
Stay tuned for Part 3: Electrical Restoration, Transistor Matching, and Tuning Results . . .
Attachments
Magnavox A531 Germanium Console Amp Reborn as a Stand-Alone Hi-Fi Component
Part 3: Electrical Restoration
Continuing from Part 1, Chassis Restoration, and Part 2, Circuit Walk-Through
If you are just joining me here, I recommend starting with Part 1 for the introduction. Germanium amplifiers are fascinating, and really aren't seen any more. I'm doing this series, and other related articles, to try to introduce some Germanium back into the fold and to serve as a resource for others who have not seen these designs before. There are still workable examples from consoles and old 8-track players that provide great fun and learning for very little cost. If you're tired of the same old amplifiers designs, maybe try some experimentation here.
Chassis Electrical Restoration
Remodeling point to point wiring will require being familiar with the schematic and circuit layout, and making sketches to determine the new connection strategy is handy. You are likely going to be replacing older larger old axial lead capacitors with much smaller radial lead versions, and you likely will not be able to reach the previous connection points. Adding bits of wire to make up the distance can work, but does not offer very good support, and may not be safe at high voltages. New terminal strips are available from Antique Electronics Supply and the like, and can be soldered onto the chassis with a larger soldering gun with some flux. You will want to plan how the new components will connect to the strips, and how they will fit safely into the existing spaces. Documenting with "before" photographs can be helpful if you take enough clear ones, but never as clearly as a good sketch.
Unsoldering crowded connections at terminal strips can be a bit challenging for even an advanced solderer, and just requires patience and practice. I typically heat the solder joint thoroughly while adding some new solder to get the old stuff to flow better, then suck as much of the loose solder out that I can with my solder sucker pump. De-soldering braid is not very effective here- It will usually not remove enough solder and you will use a ton of it. I will try to heat the joint with the iron in one hand, and use a sharp pick point to get under a loop of wire and pry it open an bit at a time. Try to resist chewing up your soldering tip using it as a hot chisel, but sometimes you just have to do it. I find a neat trick is to melt the joint then wiggle the component or wire as the solder cools, forcing a cold crumbly solder joint making it much easier to see how the wires are looped, and dig one loop out at a time. Work slowly, be patient, one layer of wire at a time, and try to do it cleanly and neatly.
I also highly recommend making a temporary work stand for your amplifier chassis. There is no reason to have your beloved chassis rocking around on top of transformers and cap cans. For mine, I cut two matching triangles out of 3/4" wood scrap and mounted them to the chassis base with some big screws and washers. This allows the amp to be tipped in two different orientations and makes for a good solid base for working and testing.
As part of the cosmetic restoration, I had already removed the balance pot, radio "control" chassis connector, phono power connector, and phenolic strip with spade connectors for the speakers and pilot light by drilling out the rivets, and filled those open holes from the inside with metal patch plates. Disconnecting them from the internal connections was fairly straight forward as some wires could just be cut at the terminal strip lugs, like the pilot light wires from the transformer that would be tied off and not used for anything else. The majority of the connections should be completely unsoldered, leaving the terminal strip lugs free for new connections in the updated circuit.
Here is a schematic of the fairly involved connectivity for those connectors, most of which was easily removed:
I kept the original power cord and entry point, but installed a new insulated in-line fuse holder and chassis mounted power toggle switch. I kept the 0.01uF capacitor across the transformer primary for noise abatement, and wired the pretty cool "TV" accessory AC outlet to be switched. Most everything here was removed along with the transformer to open the area for metal patches to be installed and cleaning, then put back in place. I had two spare terminal strip lugs to tie off the pilot light secondary for possible future use, and zip tied down the fuse holder and loose wiring to keep everything neat.
Next step was to evaluate and replace the electrolytic capacitors. I have a really inexpensive component tester that measures capacitance value, leakage and ESR. It is not precision by any stretch, but generally does a good enough job of indicating good versus bad capacitors and is easy to use. The capacitors inside the can tested more or less OK, but were leaking electrically and had fairly high ESR. Certainly they would fail in coming years. The big, black plastic axial lead electrolytics had drifted badly. In general, the smaller the physical size of the capacitor, the more electrically leaky it was. It was interesting to note that most of the capacitors measured well above the stated capacitance, some by as much as half-again high. I have observed this frequently in older electronics, and have heard that this can be a result of the capacitor drying out, which seems much more likely than being manufactured that way.
Several smaller signal coupling and decoupling capacitors were replaced with smaller, radial lead capacitors and were fit where I could, and not being all that critical because they were pretty small and light, not needing extra mechanical support. The larger capacitors are more problematic. The two black plastic 400µF at 35V speaker output coupling capacitors were very large and long, reaching over 3 inches. I wanted to replace them with radial lead 2200µF at 63V capacitors that would require a new mount location and support. The two power supply filter capacitors inside the can, 1000uF and 500uF would be replaced with 4700uF and 1000µF at 50V radial caps, so it was easiest to rip up both the output and power supply areas to make room and rebuild at the same time. In the picture below (before on the left, after on the right) you can see the changes. I soldered a new terminal strip to a ground tab above the can cap and mounted the two output capacitors tight to the terminal lugs and hot glued them to the chassis. By the way- that light green fuzz that is growing on the rivets is cadmium- a toxic metal used in the electroplating. Please be careful to wash your hands after handling a chassis that has this.
The can cap was disconnected and abandoned in place to preserve outward appearances. The original rectifier terminal strip had open lugs and a large open space in the corner, so I mounted the new power supply filter capacitors there and glued them down. The ground lug of that terminal strip had only a rivet to the chassis, and before I trusted that for a ground connection I cleaned the rivet with a wire brush, applied flux, and soldered it to the chassis for a secure connection. Don't trust chassis rivets for ground connections, as they are likely to loosen over time and give you noises and flaky connections, especially as the chassis heats up and cools down. In the photograph you can also see the phenolic strip connectors that were removed and the holes patched, the new more attractive bolts for the transformer, and the new speaker banana jack outputs at the back panel.
Multiple resistors had also badly drifted in value, and were of low tolerance to begin with (silver = +/-10%). I find that carbon composition resistors that run hot are the most susceptible to drifting. The 47Ω carbon resistor dropping the -38V to the -33V in the power supply was replaced, as were the large 2W carbon resistors (330Ω and 270Ω) used in the channel bias voltage divider string. Those had drifted as much as 20%. While I was in there, I replaced the 3.3Ω base-emitter bias resistors, which had drifted to well over 4Ω. Those were replaced with 1% tolerance metal film resistors which are now widely available in kits from Amazon for very lower cost. (There are a few new carbon resistors here and there, I use what I have . . .)
Stay tuned for Part 4: Germanium Output Transistor Matching, and Tuning Results . . .
Hi wparks,
I began my career working on this stuff. Memories.
No, 20% tolerance carbon composition resistors were standard. The 10% stuff was used in service. 5% resistors were expensive and known as close tolerance parts. If ou saw 2% part, you were working on something special, and if you needed a replacement you were waiting for it to be shipped. So, carbon composition resistors do drift like crazy as you noted. They also have high resistance changes due to either voltage across them, or heat. The pot was by far the least expensive solution and probably critical for later service.
Really nice job on the restoration.
I began my career working on this stuff. Memories.
No, 20% tolerance carbon composition resistors were standard. The 10% stuff was used in service. 5% resistors were expensive and known as close tolerance parts. If ou saw 2% part, you were working on something special, and if you needed a replacement you were waiting for it to be shipped. So, carbon composition resistors do drift like crazy as you noted. They also have high resistance changes due to either voltage across them, or heat. The pot was by far the least expensive solution and probably critical for later service.
Really nice job on the restoration.
Thank you, anatech- I appreciate the perspective provided by experience with older equipment and designs. I worked in an electronics repair shop for Sears in the 90's while getting my engineering degree, and had the privilege of seeing what happens to electronic components roasting for a couple of decades in the back of old TV's and stereos- where they cheaped out, how and why they fail and what they do. It made me a much better designer having seen these problems and knowing how to avoid them. Later, I enjoyed a hobby of restoring antique tube radios, refitting dozens of chassis full of ancient component of every style and construction failing in every way imaginable. You start to see an interesting dichotomy- on the one hand, it's amazing how often crappy stuff works at all, while on the other hand seeing just how tortured and sickly a decent circuit can become and somehow still function.
Right now I'm working up the next installment of this series on the Magnavox A531 Germanium Amplifier on Matching Germanium Output Transistors, and will be posting that soon. Thank you again. -W
Right now I'm working up the next installment of this series on the Magnavox A531 Germanium Amplifier on Matching Germanium Output Transistors, and will be posting that soon. Thank you again. -W
Thank you. I really appreciate it. Glad it was worth checking out! -W
Magnavox A531 Germanium Console Amp Reborn as a Stand-Alone Hi-Fi Component
Part 4: Germanium Output Transistor Matching
Continuing from Part 1, Chassis Restoration, Part 2, Circuit Walk-Through, and Part 3, Electrical Restoration
If you are just joining me here, I recommend starting with Part 1 at the top for the introduction and the whole journey.
Germanium Output Transistor Matching
For a germanium amplifier, replacements of the original output transistors will likely never be had unless you get lucky and find them one by one on eBay at a high price, or buy a parts chassis. The only other option may be to significantly modify the biasing to use modern silicon replacements. Therefore, you will be stuck using the transistors you have (assuming they work) and any matching will be limited to re-arranging what you have. I was in that camp, until unbelievably, inconceivably, impossibly, I actually found three of the 38P1C numbered output transistors in a Magnavox repair envelope for a reasonable cost on eBay. Since the Magnavox per-model internal secret numbering scheme used on these transistors completely hides what the actual transistor parts were, this truly a one in a million outcome. (This is why I like playing with Japanese germanium amplifiers from old 8-track players, that usually use industry numbered parts.) Having three new old stock replacements provides the opportunity to characterize the total set of seven transistors, and match to put together the best set.
Measuring current gain of germanium transistors is a little more involved than just a quick reading on a component tester. Germanium transistors have a LOT of leakage current compared to silicon, and this leakage current will be confused for current gain by the tester, giving highly varied erroneous values for gain. However, there is a very simple test circuit with just a battery and two resistors that will let you first measure leakage alone, then subtract it to obtain the actual current gain. This circuit is also useful for troubleshooting, and for testing new old stock germanium transistors that may have too much leakage to be usable in whatever circuit you are auditioning for, such as for guitar fuzz pedal circuits. In addition to what I will present here, I have found an article online that discusses this procedure, providing a supplemental perspective: "Picking transistors for FF Clones" by R.G. Keen
Below is the very simple test circuit using one 9V battery, two resistors of non-critical value, and some clip leads. If you flip the polarity of the battery it can be used on NPN devices, but germanium power transistors are always PNP, so I focus on this. The idea is to place the transistor in the "active" mode of operation, meaning that the transistor is partly on, so the operating current flowing between the emitter and collector is proportional to the base current, (plus some ~fixed leakage current). For a PNP transistor in active mode, the base voltage is lower than the emitter voltage so the base-emitter junction is forward biased, and the collector is lower still so the base-collector junction is reverse biased so current can flow out of the collector to ground. So, the emitter will be connected to the highest voltage of +9V (ish, not critical). The collector will be connected to GND through a 1K resistor, allowing us to measure output current coming out of the collector. The base will be left disconnected for the leakage measurement, then pulled partly down through a high resistance to ground allowing a small amount of base current to flow for the leakage + proportional current measurement.
Below are the details of the test circuit. It is not necessary that the voltages and resistances match the operating conditions inside the amplifier, as for matching we only need a relative comparison between different transistors under the same conditions. The linked article above and similar discussions of this testing often use a "2.472KΩ" collector resistor to simplify the gain reading, which is unnecessary if you are able to operate a pencil and calculator. Just use whatever resistance is handy in these ranges, measure and record the actual resistance, and you will be able to calculate the correct currents. I am using a roughly 1k (983Ω) collector resistor as it is more than adequate for our testing. Keep in mind these are power transistors, so the leakage current will be quite a bit higher than for the small signal transistors used in fuzz pedals, so do not be alarmed at milliamps of leakage, it's typical for the generally incontinent germanium devices.
The test procedure is simple- you will first connect just the emitter and collector clips, and leave the base disconnected. In this state, the transistor is off, only leakage current (that is independent of base current) will flow from the collector. Measure the voltage across the 1KΩ (RC) and calculate this leakage current (I = V/R). Next we connect the base clip to allow the transistor to turn partly on. We measure the voltage across the 2.2MΩ (RB) and calculate the base current. Now measure the new higher voltage across RC for the combined collector current (which is now leakage + proportional current). With those three current numbers, and can subtract out the leakage current and calculate the correct current gain = (proportional current / base current). Pretty easy, especially if you draw up a table to fill in with the three voltage measurements as you take them, and do the math to calculate the currents afterwards. Keep in mind that leakage current is =highly= dependent upon temperature. Even briefly touching the transistors with your fingers will warm them up and greatly change the readings, throwing off your gain calculations. (Just watch the voltage across the collector resistor as you touch the transistor- you will be surprised at how sensitive it is, and how much the reading changes.) Label the transistors, and arrange them legs up on your bench, ready to be tested, and let them cool back to room temperature for a couple minutes before taking readings. Try not to touch them at all as you test them, so alligator or clip lead connectors will make this much easier to do and greatly improve your accuracy.
Below is the table of measurement results for the three replacement (R) and four original (O) output transistors. The leakage varies wildly from a low of 0.58mA to over 1.60mA, and will be a LOT higher at the actual amplifier operating voltages and temperatures, but these results will still provide useful comparison. The current gain also varies widely, with a low of 35 and a high of 72. Fortunately the rest of the transistors had a fairly well grouped gain centering around 50. From the onset I will exclude the low and high gain outliers (both original) , leaving five to work with. For selecting matched devices to work together in a class AB output stage, matching the current gain of the two (usually complimentary) devices that work together within the same channel is the most primary consideration. As the output swings above and below the quiescent center point, alternating which transistor is mostly on and which is mostly off, matching the current gain of the two devices provides the best symmetry between the upper and lower halves of the amplified waveform, reducing distortion. After the devices within each channel are well matched, minimizing the difference between left and right channels is a secondary consideration, and may only be possible if you have a good pool of devices to choose from so you can put together a matched quad. Certainly not an option here, but fortunately they should match pretty well. This is where global negative feedback comes in, to reduce the effects of these mismatches even further.
A lesser consideration in matching is the amount of leakage between the devices. While the leakage current is relatively independent of the base current, it does vary significantly with temperature and collector-emitter voltage. For any decent heat sink steady-state temperatures should be fairly stable and similar across the output devices. Voltage, however, can vary significantly at high volume, as the output waveform swings high and low, varying the collector-emitter voltages. This will result in a varying leakage component that will actually slightly reduce the current gain, but generally this leakage will be small compared to the actual current gain and the quiescent bias current, so the leakage is of small concern. Where different amounts of leakage between the two devices in the same channel becomes more visible is in the quiescent output voltage. After I replaced all of the bias voltage divider ladder resistors between the two channels with 1% metal film, I still observed as much as a volt or more difference between the quiescent output voltage between the left and right channels when I installed the highest and lowest leakage devices together in the same channel. While this DC offset between the channels does not really matter (because of the output capacitors) and is not really audible (it really only changes the clipping limits) it is a minor annoyance, and more about "the principle" than anything. How much mismatch in actual current gain I am willing to sacrifice in order to more closely match the leakage, however, is "not much".
My initial selection would have been R2 (gain = 52) and O2 (53) for one channel, and R1 (50) and R3 (49) for the other. However, upon looking at the leakage, this would give an imbalance between R2 with the lowest leakage of .58mA and O2, the highest leakage of 1.44mA, and I would get that large DC offset. After much deliberation and judicious strategery, I have settled on the pairing of R1 (50) @ .88mA with R2 (52) @ .58mA, and R3 (49) @ 1.3mA with O3 (48) @ 1.0mA. This seems to me to be the best overall trade-off between both current gain and leakage- a 4% difference in gain in one channel, 2% in the other, and leakage not hugely mismatched. That is how I have installed them. The remaining transistor, O2 (53) @ 1.44mA, will still make a fine first-string replacement candidate should one fail.
Stay tuned for Part 5: Tuning and Measurement Results . . .
Last edited:
Magnavox A531 Germanium Console Amp Reborn as a Vintage Stand-Alone Hi-Fi Component
Thank you very much. I appreciate it. More to come soon.
Thank you very much. I appreciate it. More to come soon.