Hi,
I am planning to build a stereo set of the Honey Badger. The power supply will consist of a chain of four open frame SMPS that a German supplier sells for less than 5 Euro. Each SMPS is rated at 24 volts, 2.5 amps. The voltage can be adjusted from about 22 to 26 volts by a trimpot. I just tested two units, connected in series, feeding a 10,000 µF electrolytic. They worked fine and could deliver up to 2.8 amps @ 16,5 Ohms for more than an hour. The only disadvantage is, as far as I can see it up to now, that they don't manage hickup recovery. I.e., though they can easily withstand this overload, they may not start up against it. One item (not always the same one!) stays powered off. But as such a severe load hasn't to be expected from a power amplifier at start up, I think that I can build a +/- 52 volts power supply for my Honey Badgers, consisting of four of these SMPS per channel - at a budget of less than 40 Euro (!), plus two electrolytics 10,000 µF/63 volts, one per rail. Expected output power with these will be about 180 Watts@4 ohms per channel.
Some questions and considerations about the Honey Badger itself that aren't adressed in the build documentation:
1. None of the suggested transistors for Q1 and Q2 seen to be available here in Germany. So I intend to use BC550C's instead. Are they ok?
2. Is it absolutely necessary to use the same devices for Q7/Q6 and Q11/Q12, respectively? Or could Q6 and Q11 be replaced by cheaper small signal devices instead, such as BC546B's?
3. Could the Vbe multiplier Q13 be replaced by another TO-126 device, such as a cheap BD139?
4. I'd like to chose 1,800µF/63 V electrolytics for C13 and C17, the largest ones I could find here that still fit into the PCBs.
5. Are the Honey Badger PCBs still available from the DIY Audio shop here?
Best regards!
I am planning to build a stereo set of the Honey Badger. The power supply will consist of a chain of four open frame SMPS that a German supplier sells for less than 5 Euro. Each SMPS is rated at 24 volts, 2.5 amps. The voltage can be adjusted from about 22 to 26 volts by a trimpot. I just tested two units, connected in series, feeding a 10,000 µF electrolytic. They worked fine and could deliver up to 2.8 amps @ 16,5 Ohms for more than an hour. The only disadvantage is, as far as I can see it up to now, that they don't manage hickup recovery. I.e., though they can easily withstand this overload, they may not start up against it. One item (not always the same one!) stays powered off. But as such a severe load hasn't to be expected from a power amplifier at start up, I think that I can build a +/- 52 volts power supply for my Honey Badgers, consisting of four of these SMPS per channel - at a budget of less than 40 Euro (!), plus two electrolytics 10,000 µF/63 volts, one per rail. Expected output power with these will be about 180 Watts@4 ohms per channel.
Some questions and considerations about the Honey Badger itself that aren't adressed in the build documentation:
1. None of the suggested transistors for Q1 and Q2 seen to be available here in Germany. So I intend to use BC550C's instead. Are they ok?
2. Is it absolutely necessary to use the same devices for Q7/Q6 and Q11/Q12, respectively? Or could Q6 and Q11 be replaced by cheaper small signal devices instead, such as BC546B's?
3. Could the Vbe multiplier Q13 be replaced by another TO-126 device, such as a cheap BD139?
4. I'd like to chose 1,800µF/63 V electrolytics for C13 and C17, the largest ones I could find here that still fit into the PCBs.
5. Are the Honey Badger PCBs still available from the DIY Audio shop here?
Best regards!
If your SMPS holds up @ ±52Vdc during a short term loading of maximum power, then you should easily reach 200W into 4ohms and maybe 220W into 4ohms.
The current demanded by the 4ohms dummy resistor will be sqrt(220/4)= 7.4Aac = 10.5Apk.
This has to come from the 10mF capacitor and the SMPS has to recharge that capacitor during the rest of the duty cycle.
I think you have too small an SMPS @ 240VA for an ~200W into 4ohms amplifier and 4ohms speaker. If you change to an 8ohms speaker you may find the SMPS is OK.
What bias current do you intend to set your amplifier to?
That determines the quiescent current of the load in parallel to the 10mF smoothing capacitor that the SMPS needs to be able to start, or re-start when turned ON.
The current demanded by the 4ohms dummy resistor will be sqrt(220/4)= 7.4Aac = 10.5Apk.
This has to come from the 10mF capacitor and the SMPS has to recharge that capacitor during the rest of the duty cycle.
I think you have too small an SMPS @ 240VA for an ~200W into 4ohms amplifier and 4ohms speaker. If you change to an 8ohms speaker you may find the SMPS is OK.
What bias current do you intend to set your amplifier to?
That determines the quiescent current of the load in parallel to the 10mF smoothing capacitor that the SMPS needs to be able to start, or re-start when turned ON.
Thanks a lot, Andrew! Greatly appreciated! As I said before, the rails' voltage can easily be adjusted just by a turn on those SMPS' trimpots between ±44 and ±52Vdc. My considerations now are as follows: With eight of those cheap SMPS's I get a rated load capability of (8 x 60=)480 watts DC input power. Two thirds of this (roughly the efficiency of a class B amp, driven sinusoidally) means 320 watts output power, i.e. 160 watts @ 4 ohms per channel. More than enough for a living room, isn't it ;-)? According to the build advice, I intend to bias each output device to 50mA, that is 150mA in total per channel - and also per SMPS' chain. That means an ohmic load of 320 ohms per 48Vdc rail. I'll do this test soon - two SMPS in series, with a load of 10,000µF and 330 ohms (just an E12 value...) in parallel. We'll see, but I'm quite sure that nothing strange will happen. Prior to building my Honey Badger, I'll test this arrangement with two power modules I've got a year ago, salvaged by a friend from a K+H ES707 integrated amplifier. These were made about 1971, basically representing a well-known RCA design, but with some major flaws I think the designers yet didn't recognize in those days. I intend to tweak them from quasi complementary to true complementary, as we have proper PNP devices nowadays. Saying that, it better would be the topic of another thread - or the thread concerning those old RCA designs. Last but not least, what do you think on my concerns, regarding the Honey Badger itself - especially on the search for the PCBs? Best regards!
Hi,
just did the testing: 2 units of my SMPS, connected in series, easily started on a load of 10mF in parallel with two 150 ww resistors in series, that means 160mA DC current. Seems to be promising!
Best regards!
just did the testing: 2 units of my SMPS, connected in series, easily started on a load of 10mF in parallel with two 150 ww resistors in series, that means 160mA DC current. Seems to be promising!
Best regards!
Hello ostripper
For more than a month now I have been trying to design a power amp based on input jfets and output laterals.
After struggling with LT spice I decided to let go the initial idea posted here: http://www.diyaudio.com/forums/solid-state/297929-very-hq-power-amplifier-assemblage-vii.html
While searching for ideas in this forum I found your wonderfull work and would like to implement it with input jfets and output laterals.
Please let me know if it can be done this way:
For more than a month now I have been trying to design a power amp based on input jfets and output laterals.
After struggling with LT spice I decided to let go the initial idea posted here: http://www.diyaudio.com/forums/solid-state/297929-very-hq-power-amplifier-assemblage-vii.html
While searching for ideas in this forum I found your wonderfull work and would like to implement it with input jfets and output laterals.
Please let me know if it can be done this way:
feedback from R5 has to tie into the top of R4.
R25 is too low to protect the EF. I calculate 1.3W, when supply rails are 75Vdc
D1 & 2 need to be power diodes to rupture the fuse in faulty source equipment.
Add protection diodes across the NFB DC blocking cap. These can be fast 914 type.
The basic circuit looks about right.
R25 is too low to protect the EF. I calculate 1.3W, when supply rails are 75Vdc
D1 & 2 need to be power diodes to rupture the fuse in faulty source equipment.
Add protection diodes across the NFB DC blocking cap. These can be fast 914 type.
The basic circuit looks about right.
Last edited:
I changed the VBE multiplier for a resistor and made your suggested mods.... also modified the miller caps according to dadod instructions....
Also added two more mosfets and some emitter resistors...
Now I would like to use lighter/faster drivers like ksa1381 can these be used with 6 mosfets ?
Also added two more mosfets and some emitter resistors...
Now I would like to use lighter/faster drivers like ksa1381 can these be used with 6 mosfets ?
calculating R25.
The worst case dissipation occurs when the transistor impedance equals the total resistor impedance of both the collector side and the emitter side.
If you use 220r in the collector side and 820r in the emitter side then you have a total of 1040r
Add on the same again for the effective transistor impedance and you have 2080r resisting the 75V from -ve to zero volts.
P=75²/2080 = 2.7W. Half is dissipated in the two resistors and half in the transistor.
You should be looking for a total dissipation of <1.2W for a 600mW transistor.
That would require total impedance = 75V²/1.2W = 4688r
You need a total of 2344r just to make the transistor worst case dissipation = 1.2W
You have some extra due to the transient nature of the overload event. You also need to take account of worst case highest rail voltage.
I would use a bigger transistor (1W?) and a bigger resistor, maybe 820r Re and 2k Rc
But you may find that 1k6 or 1k8 works with a small transistor. Especially since many design and build without any protection/current limiting resistor in the EF helper, eg D.Self.
The worst case dissipation occurs when the transistor impedance equals the total resistor impedance of both the collector side and the emitter side.
If you use 220r in the collector side and 820r in the emitter side then you have a total of 1040r
Add on the same again for the effective transistor impedance and you have 2080r resisting the 75V from -ve to zero volts.
P=75²/2080 = 2.7W. Half is dissipated in the two resistors and half in the transistor.
You should be looking for a total dissipation of <1.2W for a 600mW transistor.
That would require total impedance = 75V²/1.2W = 4688r
You need a total of 2344r just to make the transistor worst case dissipation = 1.2W
You have some extra due to the transient nature of the overload event. You also need to take account of worst case highest rail voltage.
I would use a bigger transistor (1W?) and a bigger resistor, maybe 820r Re and 2k Rc
But you may find that 1k6 or 1k8 works with a small transistor. Especially since many design and build without any protection/current limiting resistor in the EF helper, eg D.Self.
Last edited:
Hi Andrew,
I don't understand what you say, as both in the schematics and the PCB layout/silk screen, given in the build advice, R5 yet returns to signal gnd.
Why? I've never seen, in no design, any diodes in parallel with the NFB's dc blocking capacitor.
Btw, is there really any demand for this cap to be unipolar? In a PNP differential pair input section an electrolytic should be expected, with it's anode returning to gnd.
Best regards!
Yes, I am!
But I suspect we might be talking about different schematics. My considerations relate to the Honey Badger schematics, as given in post #2. And yours?
Best regards!
I am sorry... maybe I started a real confusion... Now back to the point.
In my implementation of the HB with fets and LMOS I have perfect behaviour as long as input is equal or below 1.38Vdc.
Current in Q9 (VAS) is around 1mA and dissipation 60mW.
When I reach 1.45Vdc input, things get really nasty with current over 180mA and dissipation of 4W in Q9 (VAS).... It is the treshold of cliping at 400W output into 8ohm.
In my implementation of the HB with fets and LMOS I have perfect behaviour as long as input is equal or below 1.38Vdc.
Current in Q9 (VAS) is around 1mA and dissipation 60mW.
When I reach 1.45Vdc input, things get really nasty with current over 180mA and dissipation of 4W in Q9 (VAS).... It is the treshold of cliping at 400W output into 8ohm.
Now in the case of the original HB circuit by OS, I have almost the same behaviour.
Below 1.4Vdc input everything is perfect and when I reach 1.5Vdc input I get over 200mA on Q9 with dissipation over 3W.
Maybe it is a simulation glitch or does it mean we can not drive these circuits into clipping ?
Below 1.4Vdc input everything is perfect and when I reach 1.5Vdc input I get over 200mA on Q9 with dissipation over 3W.
Maybe it is a simulation glitch or does it mean we can not drive these circuits into clipping ?
I do not know how good LT spice is (I'm still learning it) but in real world overloading causes oscillation which can be destructive. Did you try simulation with overload protection diode (eg bav21) between Q9 base and Q10 collector?
I have some problems with entering components into LT spice libraries with my Windows 8.1 so I cannot run many circuits yet but I vaguely remember that in older LT spice on Windows XP driving HB into clipping created problems and distortion was skyrocketing. I did not measure currents here and there but probably that was happening.
cheers,
I have some problems with entering components into LT spice libraries with my Windows 8.1 so I cannot run many circuits yet but I vaguely remember that in older LT spice on Windows XP driving HB into clipping created problems and distortion was skyrocketing. I did not measure currents here and there but probably that was happening.
cheers,
Clipping of audio/music is a transient effect, until you have a drunkard in charge of the volume at a party night.
transient clipping is a very short term effect and the amplifier does weird misbehaviour during the indicent.
I don't know if simulation can replicate what happens during the incident. You have to be very clever in asking the correct questions to get out sensible answers.
I just use an amplifier that has ~ +20dB headroom over my average signal levels fed to the speaker/s. And try to find an amplifier that recovers gracefully from mild clipping.
A current limiter using resistance in the high current route of a near saturated transistor is an easy protection to implement. What you do need to do is ensure it does not ruin the reproduction of the audio/music.
transient clipping is a very short term effect and the amplifier does weird misbehaviour during the indicent.
I don't know if simulation can replicate what happens during the incident. You have to be very clever in asking the correct questions to get out sensible answers.
I just use an amplifier that has ~ +20dB headroom over my average signal levels fed to the speaker/s. And try to find an amplifier that recovers gracefully from mild clipping.
A current limiter using resistance in the high current route of a near saturated transistor is an easy protection to implement. What you do need to do is ensure it does not ruin the reproduction of the audio/music.
Why are you unbalancing the LTP with a variable resistor between the sources?
Is there an imbalance in the operating conditions of the LTP halves?
Have you checked the current injected by Q9 on one side to the current injected by the two bases of the CM into the other side?
These currents must be balanced and you need to check this during the build testing.