Hi,
I'm in the process of designing an amplifier using an emitter follower triple (EF3) output stage. This is a project that has constantly evolved over the years but never eventuated into a physical unit. That's about to change and I'm committed to building this sucker in 2022.
I'm after some advice in choosing a suitable driver BJT for this output trio. Here is some background information about the design to help with the advice.
Originally I had selected MJE15030/MJE15031 as the driver pair and then decided to change to the MJL3281AG/MJL1302AG output BJT's as the driver. However, I'm concerned about the higher collector-base capacitance (Cob) and the ability of the predriver to cope with this.
Has anyone built a similar output stage or just has some words of wisdom?
Cheers
Paul
I'm in the process of designing an amplifier using an emitter follower triple (EF3) output stage. This is a project that has constantly evolved over the years but never eventuated into a physical unit. That's about to change and I'm committed to building this sucker in 2022.
I'm after some advice in choosing a suitable driver BJT for this output trio. Here is some background information about the design to help with the advice.
- Based on the Locanthi T circuit.
- Design is 150W RMS into 8 Ohms, 300W RMS into 4 Ohms (approx).
- 3 Pairs of Paralleled output transistors (MJL3281AG/MJL1302AG) or ThermalTrak equivalent.
Originally I had selected MJE15030/MJE15031 as the driver pair and then decided to change to the MJL3281AG/MJL1302AG output BJT's as the driver. However, I'm concerned about the higher collector-base capacitance (Cob) and the ability of the predriver to cope with this.
Has anyone built a similar output stage or just has some words of wisdom?
Cheers
Paul
Can’t offer technical perspective, but Google/search DIYA for the “blameless” thread.
I have a 5-pr with those same outputs and same driver combination, so that thread should offer you some thought starters.
I think the thread was started by OStripper (designer?)
I have a 5-pr with those same outputs and same driver combination, so that thread should offer you some thought starters.
I think the thread was started by OStripper (designer?)
C5242/A1962, or C0281/A0302, Whichever you can get stock of. Good compromise between the largest outputs and MJE1503x, which can be a bit small sometimes. And TO-220’s can be a pain in the neck anyway.
Use a “VAS transistor” for the predriver, and it will be fine.
Use a “VAS transistor” for the predriver, and it will be fine.
I have the same construction 3EF as you described in my own build 6 years ago. It's of slightly higher power than yours at 200W into 8. I have 5 pairs of ThermalTrak NJL3281/1302 in final stage, and a pair of MJE15034/5 to drive them. In my PCB design and heatsink arrangement I did have provision of NJW 0281/0302 as optional driver stage. I did not find MJE15034/5 inadequate. That amp drives a pair of thirsty 4-ohm PSB Stratus Gold-i and has been on 24/7 since.
Personally I'd favor MJE15034/5 or MJE15031/2 as drivers, as their ft rises a lot sooner over collector current than do the higher power alternatives, and that can translate to a superior stability margins, at least so in simulation. The concern of them being lesser in current/power handling can be eased with an overly beefy final stage, hence 5 pairs.
Personally I'd favor MJE15034/5 or MJE15031/2 as drivers, as their ft rises a lot sooner over collector current than do the higher power alternatives, and that can translate to a superior stability margins, at least so in simulation. The concern of them being lesser in current/power handling can be eased with an overly beefy final stage, hence 5 pairs.
No Miller effect in an emitter follower, I wouldn't worry myself, but you could multiply the max slew rate anticipated by the capacitance to get the current burden of that capacitance and compare with the predriver's current level.
For instance 10V/µs x 600pF = 6mA
Note a 100Vrms sinusoid at 20kHz has max slew of 12.5V/µs for comparison.
BTW: oversizing a driver doesn't make sense when the outputs are sustained beta devices with pretty flat gain all the way to full load. 3281/1302's have about 100x current gain pretty solidly, your drivers are running around the 100mA sort of area typically.
[In the old days of 2N3055/2N2955 then the drivers could be doing nearly as much work as the outputs in extremis!]
For instance 10V/µs x 600pF = 6mA
Note a 100Vrms sinusoid at 20kHz has max slew of 12.5V/µs for comparison.
BTW: oversizing a driver doesn't make sense when the outputs are sustained beta devices with pretty flat gain all the way to full load. 3281/1302's have about 100x current gain pretty solidly, your drivers are running around the 100mA sort of area typically.
[In the old days of 2N3055/2N2955 then the drivers could be doing nearly as much work as the outputs in extremis!]
After the watt wars started, Peavey in a PV1500 used MJE15031/32 as drivers for 4 pairs 2SA1943/2SC5200 outputs. When the power ratings in the name were 4 ohms 24/7, the CS800s used TO3 drivers for 4 pairs TO3 outputs. If you are not running 150 W/ch with a high crest factor (techno, house) for 6 hour shows, you should be fine with the TO220 On drivers.
For this output power and 3EF I would use TTC011/TTA0006 as predriver, NJW0281/0302 as driver, and 4 pairs of MJL3281/MJL1302 as output.
Thanks all for your great advice, there's some really good points raised.
I forgot to mention that I'm using TTC004B/TTA004B as pre-drivers. Although I'm very fond of the KSC3503/KSA1381 as pre-drivers, as most know it's very difficult to obtained closely matched pairs due the different hFE grades of NPN/PNP in production. The TTC004B/TTA004B are better in this regard.
I started the design with NJL3281DG/NJL1302DG ThermalTrak as my outputs devices however, I have since had a change of heart and am now preferring the standard MJL3281AG/MJL1302AG with a more conventional bias scheme. I may re-visit the ThermalTrak devices in another project.
I think the biggest issue with using large power transistors as a driver, is that they have low fT at low current. The MJL3281 is around 20MHz at 100mA compared with 60MHz at 100mA for the MJE15034. The 100mA - 200mA range is about what's required to drive the outputs if I've done my math correct.
At 300W RMS into 4 Ohms I calculated the peak current at 12.25A. This equates to a total base current of 122.5mA assuming an hFE of 100 in the output devices. This tends to suggest the MJE15034/MJE15035 or the TTC011B/TTA006B would be much better suited in this position.
So I'm leaning towards TTC004B/TTA004B Pre-drivers, MJE15034/MJE15035 Drivers and MJL3281AG/MJL1302AG Outputs unless anyone can give me a really good reason not too?
Thanks
Paul
I forgot to mention that I'm using TTC004B/TTA004B as pre-drivers. Although I'm very fond of the KSC3503/KSA1381 as pre-drivers, as most know it's very difficult to obtained closely matched pairs due the different hFE grades of NPN/PNP in production. The TTC004B/TTA004B are better in this regard.
I started the design with NJL3281DG/NJL1302DG ThermalTrak as my outputs devices however, I have since had a change of heart and am now preferring the standard MJL3281AG/MJL1302AG with a more conventional bias scheme. I may re-visit the ThermalTrak devices in another project.
I think the biggest issue with using large power transistors as a driver, is that they have low fT at low current. The MJL3281 is around 20MHz at 100mA compared with 60MHz at 100mA for the MJE15034. The 100mA - 200mA range is about what's required to drive the outputs if I've done my math correct.
At 300W RMS into 4 Ohms I calculated the peak current at 12.25A. This equates to a total base current of 122.5mA assuming an hFE of 100 in the output devices. This tends to suggest the MJE15034/MJE15035 or the TTC011B/TTA006B would be much better suited in this position.
So I'm leaning towards TTC004B/TTA004B Pre-drivers, MJE15034/MJE15035 Drivers and MJL3281AG/MJL1302AG Outputs unless anyone can give me a really good reason not too?
Thanks
Paul
Certainly no reason *not* to. “Bigger” drivers would only offer advantage if you were planning on 30 to 40 amps peak current (60V supplies, 2 ohm load) as being “normal operation”. For the predrivers, I tend to use whichever of that general class are on hand rather than waiting for something specific to come in (unless voltages dictate something specific). HFE mismatch is hardly felt there.