I COMPLETELY agree with this. And not just for "improved bass". The quality of the crucial midrange improves significantly also.
Is TIP3055 (TO-3P/TO-247) suitable for this amp? I know 90watt dissipation is plenty but still want to know. Which one is better, 2n3055 or TIP3055?
But how to measure hfe at a constant current? I only have a DMM. Yes, TIP is easier to install but i've some good indian 2n3055(Ft~2.5mhz min). I've total 8 To-3 , so i can use 2n3055 for power supply also. Actually i need 2 piece of thick L type aluminium bracket. Let's see.
You should't need to measure hFE but you may get some better understanding from reading the Project 15 article on cap. multiplier supplies for class A amps at Rod Elliott's ESP site. You get the whole story and good advice there. Plus, that's where you find the original Class A amplifier site: https://sound-au.com/project15.htm
I already did & now i'm fully ready to build this amp. Just waiting for heatsink & transformer. Regards.
Last edited:
Anyway my choice of components are~ 2n3906, BD139-16, 2n3055. 2.2uf polypropylene as input cap, 470uf as bootstrap as well as feedback & 4700uf as output cap. All resistors are metal film except R1 & R2.
If you have a dual power supply of say +/-5V or two 5V supplies and a selection of resistors you can measure the gain of a transistor with a DMM or other meter relatively easily. Suppose 2N3055 at 1A. Vbe is going to be about 0.8V, so an emitter resistor of 4.2 ohms for a PSU of 5V (you can use 3.9 or 4.7 depending whether you prefer slightly under or over 1A, and if it is to check matching then not too important).
Connect the collector to +5V, the emitter resistor to -5V and put the DMM between common (centre) and base. That will give you the base current. You can match devices quickly or use a 1 ohm resistor on the base and measure voltages across both to obtain the currents. Several options!
But note the device needs to be on a heatsink and dissipation will be in the order of 5W so the resistor needs to be that as well. Also make the measurement quickly to grab the figures before it heats up too much as the temp will change the Vbe and hfe.
Connect the collector to +5V, the emitter resistor to -5V and put the DMM between common (centre) and base. That will give you the base current. You can match devices quickly or use a 1 ohm resistor on the base and measure voltages across both to obtain the currents. Several options!
But note the device needs to be on a heatsink and dissipation will be in the order of 5W so the resistor needs to be that as well. Also make the measurement quickly to grab the figures before it heats up too much as the temp will change the Vbe and hfe.
Like this? I think i should use lm3xx type regulators to build a symmetrical supply because here the current is quite high or 7xxx type regs would suffice? I already hve those in my possession so no problem. thank you btw🙂
Last edited:
Project177 of Rod elliott has great info about this(hfe test at a constant collector current), i completely missed that :-(
Last edited:
I haven't seen Rod Elliotts version but the diagram you drew was what I suggested. You could arrange for a second emitter resistor to the + supply to measure PNP's with suitable connections. You will need something like a millivoltmeter to measure the base current (voltage drop across 1 ohm). I have found an FET input op amp ideal as a pre-meter amplifier, although I have also got a discrete version which works well too.
What you suggested is exactly same as proposed by Mr.elliott. Same thing. Anyway i want to thnk you again for this simple but effective circuit.
Hello,
I build the first channel circuit and i just test it before doing the second one.
i successfully set the dc offset and bias, but when I checked the power supply voltage I saw that the voltage dropped from 25V when nothing was connected on it, to 21V.
I have 1 power supply for each channel using capacitance multiplier (https://sound-au.com/tcaas/jlhcapmult.htm) and to 33000µF instead of the 4700µF
is it expected behavior?
regards
max
I build the first channel circuit and i just test it before doing the second one.
i successfully set the dc offset and bias, but when I checked the power supply voltage I saw that the voltage dropped from 25V when nothing was connected on it, to 21V.
I have 1 power supply for each channel using capacitance multiplier (https://sound-au.com/tcaas/jlhcapmult.htm) and to 33000µF instead of the 4700µF
is it expected behavior?
regards
max
You need somewhat more powerful transformer if you want to maintain supply rails at +25vDC but 21v isn't bad for moderate listening level & 6-8 ohm load.
Let's add a little science to the theory. The mechanism at play here is known* in electric guitar circles: as the voltage sags there's a series of recharge spikes at a notional 100Hz (or 120Hz) on the power rails which intermodulate the signal, providing side bands around your actual signal. Not only at 100/120Hz but over the full spectrum of the charge spike.
In guitar amps (e.g. your Fender Bassman) this is a positive outcome. In HiFi amps this is a disaster.
This effect is almost completely removed by Class A operation but as the JLH is not balanced (that is, the power supply capacitors are involved in maintaining voltage) it's not completely removed. As the JLH is not fully balanced/differential it's PSRR is actually not that great.
Now, 99.99% of "hi fi" amplifiers use the same, fundamentally broken** power supply architecture. But it isn't the only architecture out there: my JLH has a regulated power supply (per JLH 1996 design, I believe); other amps (like Hirage's Monstre) use batteries and proper ClassA valve amps (e.g. Brooks PP 10A or Olsen's Karna) use choke input or multistage supplies (because, once upon a time, inductors were cheaper than big caps).
* see https://www.ampbooks.com/mobile/classic-circuits/class-AB-ripple/ actual reference Thomas Schmidt and Henry Westphal, "The WILDCAT Low Noise Power Supply," Final Project Report, Project "WILDCAT Guitar Amplifier," (Berlin: Technical University Berlin, 2007). (in German) linked therein
** broken in that any increase in C (to drop ripple) just results in increased ripple current but in shorter spikes, which just results in even worse noise waveshapes.
Yes, but what Rallyfinnen and I are referring to is not power supply filtering. It is the input, bootstrap, and output caps. Especially the latter two. For those, the standard values are way too small. I'd start with 8200uF in both positions, and evaluate from there. Unless you lay out your own pcb to accomodate them, the limiting factor will likely be mounting room...
Last edited:
Yep, doing some bode plots of the voltages on the caps in sim will show this. Output cap is trickier to simulate, that can be done in Xsim or similar with the actual measured impedance of your speaker.
It's easy enuf to do in your mind. We'll assume the load is a dynamic speaker, which will have wildly nonlinear impedance at low-mid freqs. Most JLH kits come with 2200uF output caps. The impedance of a 2200uF cap at 100Hz is 0.72 Ohms. At 50Hz, twice that. Series impedance linearizes phase. Do you want the amp feedback to be sensing the load through a variable series impedance, in a region where the load impedance itself is already nonlinear, and changing with amplitude as the coil moves? I think not. For output caps, the larger the better. Even with higher-impedance loads. For commercial kits, physical rooom on the pcb will probably be the limiting factor.