|17th September 2005, 03:10 PM||#91|
diyAudio Moderator Emeritus
Sorry, I don't realize that everybody use Protel
Here's the schematic, I just draw it 5 minute ago. Sorry about the quality, I got it with digital camera, not scanner.
I've used miller cap 100pf x 2, 27pf feedforward cap, it still oscilates (not with TIP35C-36C).
Maybe some of you will notice the strange part in the current mirror, I use very big 2k2. First it was 470ohm, but putting 2k2 there seems help.
|17th September 2005, 05:13 PM||#92|
Join Date: Nov 2004
If I can help you with your CFP amp. I like your current mirror and symetrical design.
I built a CFP output amp, and it's stable. It was designed for subwoofer, but in testing it plays all frequencies clearly.
Has 5 pairs of MJL4281 MJL4302 output transistors, with MJE15034/35 drivers.
I had to take some extra steps for stability, because when first built, it oscillated. Now it runs fine. The schematic shows these changes that removed the oscillation.
Added 1nf caps for B-E of drivers, and 100ohm base resistors for drivers. The 100ohm+1nf is to filter RF.
I added 4.7 ohm emitter resistors for drivers for some degenerative feedback.
Going to a 0.22uf zobel cap instead of traditional 0.1 got rid of all tapping sound, when I would touch speaker wires or heatsinks with my fingers. Also added extra 1ohm/10nf zobel to make up for long wires, and to add extra filtering.
Added 0.15 ohm output resistors (helped a ton with bias)
Added 0.27uf 250V ceramic caps to filter power supply rails and also 2KV 10nf caps from computer PSU to filter incoming DC to the capacitors.
BTW, good luck with your amp.
You may get some replies to go to EF instead of CFP but with some tweaking, it works great if you can stop it from oscillating, so keep on it! I think collector drive is stronger anyway, and I like the sound of CFP.
|18th September 2005, 03:33 AM||#93|
diyAudio Moderator Emeritus
Nice design you have. If there is "for subwoofer" and "for fullrange" design, I would choose #91 as subwoofer and #92 as full range. Because #91 will inherently have low DC offset, and with #92 I can tailor some "Niceness" (only if you want to).
The PCB that I built is not fixed pcb, the front end (Differential+VAS) is in small PCB/card, so I can try various topologies with the same main pcb.
For the output stage I choose CFP, because it is something with it's own local feedback+current drive mode, different from EF (although I'm confused now with it's oscilation ). It have big current supply capability, but not with the expense of raising RE drop or raising VBe value drop. This makes VBEmultiplier value is less disturbed from the turn-on-turn-off point of view.
I advoid using RE (like 0.15/10W in your #92), because although it will give better bias stability, it will boost the transistor turnon-turnoff characteristic, that's because I only use 1 pair for 1 amp, trying to not have parrarel. You can see speaker load can dip to 1 ohm or less (although only for u-second), that means the drop in this 0.15 will exceeded far from VBE multiplier value. The drop in this 0.15 will take more voltage drop than VBE multiplier preset value. The transistor will be hardly turn-on or turn-off, and this process will take time to settle (Like EVA said in the earlier page of this thread). I think this is also why some designs like Electrocompaniet also don't use this RE, unless we incorporated smart/sliding bias, or use many parrareled output devices (you used 5 parrarel, that's better in the view of transistor turnon-turnoff characteristic), because when sudden peak current, it is divided for 5 x 0.15ohm RE, making drop in each is less.
I've got a different card that is using the topology you have in #92. It is nice to use bootstrapped VAS, it gives more swing and more importantly it is giving very tight AC relationship between output node to VAS (due to bootstrapped capacitor), giving better audio phase response for the whole amp.
The difference in my card and your #92 is that for tail CCS I use 5k1 (like #91) supplied with carefully regulated -12V (opamp based regulator+capacitance multiplier to give very clean -12V).
Another difference is that the input and feedback Rdivider is symetrical (like #91) and I dont use LPF to ground (100pf in your #92), but using 470pf bridging from base of left to base to right (like #91). For me this method gives better result than making LPF towards ground, this 470pf will bypass any signal higher than its cut frequency not entering the differential system, and plus a bonus, this cap will again give better phase corelation between input and output node. This method is used by Electrocompaniet and suggested by JonathanCarr (well, that means it comes from 2 reputable sources).
In my card (that similiar to your #92), the OFFSET ADJ VR 680ohm is not used for adj offset, but it is used for "NICENESS KNOB" . You can try this, adjust that 680VR until the DC offset is about 100mV (don't worry, I think it won't ruin your speaker), you will be able to hear "niceness sound" (with the expense of DC offset).
This "NICENESS" knob is trendy now, look at NP's PLH
Another difference is that the value of the VBE multiplier capacitance is only 10nF (like #91), and I try to flatten the OL response by adding 220kohm // with 100pf miller cap.
In my design #91, I try to flatten this OL gain by different approach, by putting IN4148 in the differential leg. This will reduce OL loop and extend OL bandwith.
But for subwoofer amp, maybe you can put only R for differential leg, cause subwoofer amp usually need high OL gain, not care about OL bandwith.
To give more tight relation ship and better input-output phase response, I add 4k7 resistors between VBE multiplier and output node before the base stoppers (Like Threshold statis 2-3). This also gives better audio result.
|20th September 2005, 05:58 AM||#94|
Join Date: Mar 2004
Re: Speaking of Diamond buffers...
I do not know if I am late?
or did I brake the law?
but please look at simulated circuit.
Sorry for the mess, I was hurry:
The transistors of the first stage are BC546A/556A,
the second BD135/136
the third 2N3055/2955 (for the simulation purposes only).
Instead of ccs I used large resistors.
The circuit with an input 2V~ (straight from CD output) produces +- 6V~pp ath the output.
the second harmonic is at about -60dB
the third -40dB
rest at about -70dB, odd are higher.
Any sugestion or comments how to modify the values of the parts?
|20th September 2005, 10:08 AM||#95|
diyAudio Moderator Emeritus
That cct is protected by patent #6,882,225.
From the manufacturer's website, I read something. That cct likely will have DC offset bias wandering from -0.6V to +0.6V (range =1V2).
What makes this "not stable" DC offset value? All openloop diamond buffer will act like this?
|Thread Tools||Search this Thread|
|Thread||Thread Starter||Forum||Replies||Last Post|
|newbie with input tran. question and wiring question||imo||Solid State||0||18th January 2006 10:10 PM|
|Capacitor Question /question on cap||TunaFish||Parts||14||12th January 2004 02:23 AM|
|PS question - tformer question actually||breguetphile||Chip Amps||2||30th May 2003 01:29 PM|
|old speaker question, bullet midrange question & link to speaker project||wallijonn||Multi-Way||10||5th November 2002 06:03 AM|
|New To Site?||Need Help?|