The metal case is part of the circuit, usually the collector.
I don’t know if you find videos useful, but I find a couple of YouTubers really great. In particular, “xraytonyb” had just recently posted a video on how to look for transistor replacements, but it has all sorts of interesting nuggets along the way.
He has a huge back catalogue of stuff as well; everything from just soldering and chatting to how to do really complex measurements with an oscilloscope.
The best part is that you can stop it and come back later, so if you find yourself getting overwhelmed with info and details, you don’t have to keep watching. You can always come back to it later.
ahankinson posted while I was typing: Yes, the collector is connected to the metal case. This particular type of case is called TO-3, a bit outdated by now I believe. More info here: https://en.wikipedia.org/wiki/TO-3
Complex but still manageable tasks on subjects of interest, where the desire to learn/complete is very high, help to stay focused. I experience similar while programming.
On topic: @Prognosis2811 have you considered the honey badger and the like? It's same topology as the design in post #1, but more matured.
I've reviewed the honey badger schematics and examined numerous other designs as well. I'm in search of a somewhat elusive goal—a creation with an output of around 50 to 100W, streamlined without variable resistors for bias adjustment, and featuring the fewest possible components, ideally utilizing BJT, Mosfet, lateral, or Darlington transistors at the output.
While I've explored renowned Pass projects, they tend to focus on class A and lower power outputs, boasting exceptional design efficiency with minimal components.
Regrettably, my current electronics expertise places me at the base of a steep learning curve, hindering my ability to develop a circuit that encapsulates these desired attributes. My intention is to avoid integrated circuits, as I'm aiming to experience the essence of the celebrated discrete amplifiers that have garnered attention
The Apex X6 and X11 designs caught my attention during my search. While they don't quite meet my desired power output, they do exhibit the virtues of simplicity, with a modest component count and absence of trimmers. The straightforward nature of their diagrams is particularly appealing. Nonetheless, I remain receptive to any recommendations you might have in mind.
I'm open to exploring class A amplifier options, as long as they don't involve any trimmers. I understand that trimmers are essential for enhancing performance and stability, but my intention is to find a design that avoids their use while still delivering satisfying results.
While I've explored renowned Pass projects, they tend to focus on class A and lower power outputs, boasting exceptional design efficiency with minimal components.
Regrettably, my current electronics expertise places me at the base of a steep learning curve, hindering my ability to develop a circuit that encapsulates these desired attributes. My intention is to avoid integrated circuits, as I'm aiming to experience the essence of the celebrated discrete amplifiers that have garnered attention
The Apex X6 and X11 designs caught my attention during my search. While they don't quite meet my desired power output, they do exhibit the virtues of simplicity, with a modest component count and absence of trimmers. The straightforward nature of their diagrams is particularly appealing. Nonetheless, I remain receptive to any recommendations you might have in mind.
I'm open to exploring class A amplifier options, as long as they don't involve any trimmers. I understand that trimmers are essential for enhancing performance and stability, but my intention is to find a design that avoids their use while still delivering satisfying results.
I did watch the video, and I found it extremely satisfying. The author's concise and engaging manner of speaking made it far from tedious, getting straight to the point. Thank you for the excellent recommendation.
However, this has sparked a lingering question. Is Hfe not crucial when searching for a substitute transistor? While I understand the significance of matching voltage, current, power dissipation ratings, and frequency transition, does the DC current gain need to be precisely matched as well to maintain resistor and capacitor values? I'm hopeful that someone might be able to provide clarity on this specific topic.
However, this has sparked a lingering question. Is Hfe not crucial when searching for a substitute transistor? While I understand the significance of matching voltage, current, power dissipation ratings, and frequency transition, does the DC current gain need to be precisely matched as well to maintain resistor and capacitor values? I'm hopeful that someone might be able to provide clarity on this specific topic.
Add me to the list. I have the Aspergers traits'. It has been both a blessing and a curse (social phobia).
The positive effects are problem solving and a desire to experiment, as well as a tendency to ignore the rules.
Have you overlooked the need for heatsinks for the output transistors? I fiind it simpler to mount flatpack transistors such as the 2SC5200 & it's complement, the 2SA1943 on a finned one.
Yes, but that gets to the level of application in a specific circuit, rather than “will this operate at all”. In a lot of applications the hFE of a transistor is irrelevant; it’s only in specific cases where it might be critical.
And anyway, with the ability to filter these things on digikey you can simply add that as another search parameter if you feel it is necessary.
An important part of BJT design is make the circuit as insensitive as possible to hFE, as it can vary so much between devices (of the same type), and with current level. But there are critical places where the circuit performance depends entirely on high hFE, such as low noise amplifiers and a VAS transistor(*), and it matters to select particular gain bins for the device, or often to select-on-test.
Sometimes device-matching is important, you want two devices to have similar hFE and Vbe, the exact values being irrelevant.
So to be sure when substituting its best if you understand what the circuit in question expects, or else be prepared for some trial and error and disappointment.
(*) This is an application where the power-gain is desired to be extremely high - A VAS might take a signal of a few mV and 10's of µA and raise to 100V and several mA in one device - that can be around 80dB of power gain, plus a requirement of high bandwidth and high voltage - very few devices fit this spec well.
Sometimes device-matching is important, you want two devices to have similar hFE and Vbe, the exact values being irrelevant.
So to be sure when substituting its best if you understand what the circuit in question expects, or else be prepared for some trial and error and disappointment.
(*) This is an application where the power-gain is desired to be extremely high - A VAS might take a signal of a few mV and 10's of µA and raise to 100V and several mA in one device - that can be around 80dB of power gain, plus a requirement of high bandwidth and high voltage - very few devices fit this spec well.
Initially, I believed that this specific package would provide a more straightforward solution for thermal management, given its larger space. However, my perspective has shifted as I now realize that these packages are also showing signs of being outdated.
My circumstances lead me to question everything, including my own abilities, each time. Consequently, I tend to feel dissatisfied with my accomplishments, whether they are good or bad, and I often find myself starting anew from scratch. This instance is no exception. I've erased all previous work and am currently in the process of redesigning everything from the ground up. Interestingly, in this iteration, I'm substituting the output transistors with the Toshiba 2SC5200. I did find it peculiar that this design employs only the 5200 and not the complementary pair.
Last edited:
Asperger's has been quite challenging for me, particularly since I discovered that I'm on the autism spectrum alongside having ADHD. It often feels like a burden, intensified by the fact that my wife dismisses the possibility of my being autistic (despite my certainty), leading to ongoing imposter syndrome struggles.
Admittedly, understanding the cause behind my perpetual sense of being different provides some solace. However, the Asperger's aspect still lingers in the background, casting a shadow that leaves me fatigued each day. I find myself compelled to feign normalcy so that my true self remains concealed from others. Thankfully, I've come to embrace my role as a social outlier, which paradoxically reduces the exhaustion that comes from masking my true nature.
While I make an effort to adhere to certain social norms, my consistent transgression involves traffic lights, particularly when I perceive there are no cameras or other cars nearby. This behavior might deviate from the norm, yet it offers a respite from the constant masking and fatigue I experience
In the context of this specific AX6 design, would concerns about hFE not pose significant obstacles?
The FH9HVX has no trimmers, is relatively simple, sounds great by many accounts, and is well documented.
If you’re game to try class A, then the M2X is right up your alley. It’s similarly uncomplicated, fabulous sounding, sans trimmers (clever auto bias), and with plenty of documentation. Plus, there are 9 input boards to experiment with to hone your particular preferences for diyAudio sound!
You would find lots of knowledgeable people willing to help if you decide to build ether (or both, as you will likely discover once you start building these contraptions).
If you’re game to try class A, then the M2X is right up your alley. It’s similarly uncomplicated, fabulous sounding, sans trimmers (clever auto bias), and with plenty of documentation. Plus, there are 9 input boards to experiment with to hone your particular preferences for diyAudio sound!
You would find lots of knowledgeable people willing to help if you decide to build ether (or both, as you will likely discover once you start building these contraptions).
Dang. I was mostly wrong about the trimmers. 
The M2X, if built as like the original M2, can be without. The FH9HVX has a trimmer pot. Sorry for the diversion.
The M2X, if built as like the original M2, can be without. The FH9HVX has a trimmer pot. Sorry for the diversion.
After conducting an exhaustive exploration of various Apex projects, the FX4 undoubtedly stands out as perhaps the simplest configuration I could come across. This streamlined design boasts an absence of trimmers, operates on a single supply, and exhibits a modest component count: a mere 14 resistors, 10 capacitors, 4 transistors, and a lone diode. It's unlikely that I will stumble upon another schematic that is as elegantly minimalistic as this.
I'm now eagerly seeking your insights on the following inquiries:
EDIT:
I omitted one crucial detail: Are the designated output transistors well-suited for this particular project? Your discerning judgment on whether the chosen output transistors harmonize effectively with the intended functionality and performance benchmarks of the FX4 design would be tremendously enlightening. Your professional insights are eagerly anticipated and deeply appreciated.
I'm now eagerly seeking your insights on the following inquiries:
- What power supply voltage is optimal for this application? Would 35V be a suitable choice?
- Regarding its potential output capability, could it conceivably deliver 50W into an 8Ω load?
- With respect to resistor power ratings, is the general standard 250mW for most components, except for specific exceptions such as R12 and R13 which require 5W ratings?
- Perhaps most critically, if all these components were to be integrated onto a single PCB, what would be the likelihood of encountering thermal issues resulting in malfunctions within the first 10 seconds of operation?
EDIT:
I omitted one crucial detail: Are the designated output transistors well-suited for this particular project? Your discerning judgment on whether the chosen output transistors harmonize effectively with the intended functionality and performance benchmarks of the FX4 design would be tremendously enlightening. Your professional insights are eagerly anticipated and deeply appreciated.
I am certainly no expert nor professional, but where the original calls for 2sk1058 and 2sj162, those are what's called "lateral" MOSFETs, which have differences than your substitute IRFPs, which are usually called "vertical" MOSFETs. The loose nomenclature deals with how they are/were manufactured. Direct replacement of laterals with verticals usually isn't recommended without additional changes to the circuit. Other more knowlegable members here can tell you more. Current replacements for the laterals in the APEX schematic are available from Exicon. Lots of info on the site here for that.
Been out mowing my summer camp acreage. Sorry for the delay.
I have a common return to the drivers and output transistors, connected to speaker ground and power supply negative. I have a common return to the input transistor and VAS (the 2nd one). I connect those two to the output section through a small resistor, half or 1 ohm.
Here is where I buy bare Nema CE sheet. $16 for a 12"x24" piece. I drill as necessary, and use wires for point to point build. https://www.mcmaster.com/8491K7
My Apex AX6 has a single 69 v supply, produces 72 w into 8 ohms. Base AX6 has a 60 v supply for 50 watts. AX11 has a +-35 v split supply, with speaker connected to the center. Advantage of AX11, it has the constant current BC546 to give it some supply hum rejectioin. Disadvantage of the AX11, it has direct connected output transistors that, if you make a mistake like a bad solder joint, dump the entire supply current through the speaker coil. You can address this "problem" by buying a 7 to 12 part "DC on speaker protection" circuit. Most of these are snake oil, with a relay not designed to break big DC currents.
I handle the supply rejection problem of AX6 by regulating my supply, 72 v open circuit, regulated to 69 v with a series pass regulator on a separate board. Similar to the regulator built in to my dynaco ST120 chassis, but not relying on the limited gain of a 1967 2n3055 to regulate the supply current at the same time (which could not be replicated in 1985 when I bought the burned transistor ST120).
My AX6 would produce 24.3 vac on an 8 ohm speaker, for 5 seconds. That is 73 watt on 8 ohms. Same voltage out on 4 ohm speaker would be 147 watts. I don't believe a MJ15003 (which I have) would stand that current for very long.
100 w bulb series the AC line during test prevents faults from blow up parts so bad the TO3 tops bounce off the ceiling (I've done that). Slows down damage so you can probe with a meter and find out exactly what is wrong.
The parts you listed for the AX11 should work for the AX6. I think BD139/140 are little light duty for drivers, I use MJE15028/29 on my AX6 with heat sinks. TO3 output transistors are tough, but $10.50 each these days. MJL4281 are about as tough and $5 each, and only one hole to drill and get lined up, not three. 2sc5200 are as easy to mount but are not as tough as the On semi parts. I used MPS8099 as input transistor and the equivalent to MJE15028 as VAS. In a amp circuit, gain is set by the resistor ratios, most modern parts have high enough gain to meet the requirements Mike Slovatnic puts on them. On semi parts like MPS8099 (or MPSA06/56) are very high gain. I measured 300. Emitter followers like output transistors, gain is not important after the first generation parts like 2n3055 which had gain 5 at 3 amps ( horrible these days).
If you get your output bias idle current right with stacks of diodes, you can omit the bias setting pot. Pots are a reliability problem after years, the wiper oxidizes and loses contact. If you are off you can use a parallel resistor with the heat sink diode stack. Just the idle current is sensitive to what actual parts you use, no cookbook design is going to have right idle current for any combination of parts over the life of the part numbers. BTW I ended up with 2 1n4148 diodes and one shchottky diode (0.3 v drop) in the temperature sense position of the AX6. Clamped with a parallel pot, but if the pot opens up idle bias current creepts up to 40 ma, not too dangerous,.
Have fun.
Last edited:
I echo your comments and have wondered why I was different from my school friends since age of about 11. My parents decided never to tell me and I was about 40 before reality dawned.
I also have mild OCD - probably this was ineveitable, helpful even, since I work for myself running a mailorder business which is ideal for an Asperger guy since there is no face to face contact with customers.
I recently heard on the news, a hearing condition which is quite common amongst the Autistic and which I had noticed when a student. Where there is a lot of background noise, eg parties, pubs, clubs etc, I cannot understand what is being said. I just hear unintelligble speech and noise and others think I am partially deaf. It's also a problem with 'action movies' on the streaming services because the dialogue is masked by the excessively loud sound effects.
It's so stressful trying to be 'normal' and as I am in my late 60s, too old to keeping trying. Don't fight it is my philosophy - you are what your are.
I think there are many sufferers who haven't been diagnosed and perhaps think it's not a condition that afflicts them if it wasn't diagnosed during childhood.
12 v power supplies are generally used in car audio with 1 or 2 ohm speakers. In general, 1 and 2 ohm speakers sound like ****. Great for buzzing the parking lot and making the driver of the next car crazy, but not for sounding like any wooden instrument. Not high fidelity. 4 ohm speakers can be done with 40 v power supply, but AX6 as shown is best for 8 ohm speakers. 60-70 v single supply is generally used, 50 will work at reduced power. A variant of AX6 could be used at 4 ohms, that is using TO220 drivers like MJE15028/29, TO220 VAS to drive them like MJ15028, and two pairs output transistors like MJL3821 or MJL4821, or even 2sc5200. All drivers & VAS in the 2 output transistor variiety should have small heat sinks on them.
Toshiba no longer makes the 2sc5200, all sourced product are from copy artist *****, except the ones from fairchild (div On semi) & ST. On & ST fab are also in ***** but quality seems to be under control if the parts are sourced from an authorized distributor like newark, mouser, digikey. Some copy 2sc5200 on the jolly roger market are good, some copies are ****.
Toshiba and Hitachi also no longer make 2sj162 2sk1058 lateral fets, darling of ultrasimple amps for their extreme linearity. Alfet makes a copy ALF08N20 ALF08P20 but are difficult to source.
2n3773 as noted previously is a popular TO3 npn transistor, but since all ON TO3 packages are >$10.80 now you may as well go ahead and use MJ15024 which has the same die and has a guarenteed tested SOA spec. Other manufacturers like ST make the 2n3773. Note your board shown post 20, has no provision for the heat sink, which is key for output transistors producing 20-75 watts per pair. A 15x10x2 cm heat sink with 6 or 8 fins has to be mounted at the edge of the board with no little components between output transistors and the heat sink. Some people use a 1/8" aluminum angle under the TO3 packages which screws to the main big heatink. Use heat sink compound, and mica or silicon washers under the TO3 package. Triple hole drilling of the heat sink for TO3 is not something I am competent at, the drill bit in smaller sizes tends to walk sideways from the marked position.
Other popular plastic package ON NPN in that wattage class include MJL15094 MJL15096 MJW15094/15096, MJW3281 MJW4281.
As far as what wattage you need, depends on sensitivity of your speakers & listening habits. I have 98 db 1w1m speakers, listen to them mostly at 1/8 watt, 1 v average, but when the cannon goes off in 1812 overture, I can stand 50 w in my music room. 80 db 1w1m acoustic suspension speakers would require more like 2 w at orchestra pianissimo and 100 w at 72 db higher. If you're a house or techno listener, music with low crest factor, you may want 100 w all the time. In that case buying a PA amp with that kind of power, a huge heatsink, and fans, would fit your listening habits better. Huge heatsinks cost $$$, I've found the cheapest way to get them is blown up bar band amps (PA).
Last edited:
Could the configuration of these components under Minos effectively function? Considering potential alterations to resistor and capacitor values, has anyone undertaken the bold feat of integrating these intricate elements onto a sole circuit board?