Hi,
I got a question on my HB build. When I first turn it on it's dead quite, then after a minute both channels develop a hiss, then after a few minutes running when it warms up a bit the hiss dissappears. Does anyone have any idea what I should look for ? Other than this is sounds really nice once it's warm.
Thanks,
PJN
I got a question on my HB build. When I first turn it on it's dead quite, then after a minute both channels develop a hiss, then after a few minutes running when it warms up a bit the hiss dissappears. Does anyone have any idea what I should look for ? Other than this is sounds really nice once it's warm.
Thanks,
PJN
Interesting... I wiil try it.
OS, you advocate to use low current VAS on 3EF, now you change your mind?
. In DIY, we must matching something to get higher performance. Will anybody make matching hFE and VBE circuit using microcontroller? Display using 2x16 LCD, ex. "VBE 20%" or "hFE 10%".Right now, I must write vbe and hfe value for each transistors
when I match them.
No , I did not change my mind ??
This is a EF2. With a hefty load from a crappy speaker ,a mA of current
might be required to drive the MJE's.
Not the case with an EF3 - the predriver's do the job .... needing just
pA to uA's from the VAS .
An EF3 could get away with a 2mA VAS easily.
Raising VAS I on an EF3 vas might even increase THD (device dependent).
Bottom line - the EF2 is not an EF3.
EF2 = current gain of 1K-5K
EF3 = gain of 50-100K+ ..... big difference.
PS- simulate Ib for the predrivers (40uA-ef3 vs. .5-.7ma EF2 drivers)
OS
Last edited:
2SC4793 / 2SA1837
Mouser has 5K each of these in stock.
A whole 87c apiece.
These drivers might be-
1. a little closer matched
2. Higher Hfe (would not load the VAS as much)
3. A widely used driver , my HK680 uses them (for 21 years).
This was the main reason behind C18/19 ,this B-C shunt compensates
for High Ft ..... this driver pair is in my original BOM - as another
choice.
OS
Mouser has 5K each of these in stock.
A whole 87c apiece.
These drivers might be-
1. a little closer matched
2. Higher Hfe (would not load the VAS as much)
3. A widely used driver , my HK680 uses them (for 21 years).
This was the main reason behind C18/19 ,this B-C shunt compensates
for High Ft ..... this driver pair is in my original BOM - as another
choice.
OS
No , just grease 'em up and bolt them on.
My KSA/C 1381/3503 are all plastic ,
they are matched "F" grade (hard to find now) ... all plastic , bolted to the
same 35mm X 70mm piece of aluminum flashing that the original badgers
used (5 years ago)
.OS
No , just grease 'em up and bolt them on.
My KSA/C 1381/3503 are all plastic ,
they are matched "F" grade (hard to find now) ... all plastic , bolted to the
same 35mm X 70mm piece of aluminum flashing that the original badgers
used (5 years ago)
OS
Got all of those mounted on their little heatsink with thermal grease. Finished up the majority of my case, wired up my power supply and soft start and turned it on (Without amp boards hooked up) for the first time today. I wind up with +/- 58V with no load. I am using 24000uF/rail for each channel.
I did look up my DMM specs for hfe testing and since it only uses 10uA, that is most likely why my drivers were so far apart. When measuring another different complementary set, the PNPs were again much higher than NPNs with this meter. I did still order a couple more choices for drivers and they will arrive Tuesday. Hopefully I can get a reasonably close match. I will probably put together a testing circuit though since my meter's readings are not really useful for these larger transistors.
I also ordered the recommended small electrolytics mentioned earlier as well as some low ESR caps for the 1000uF and 220uF on the board because... well why not!
The next day after ordering, I found that I needed a m7 X .75mm nut for my volume pot and nobody around here carries anything like that. I can't find any 7mm, let alone a fine thread one. You would think they could include a 2 cent nut with the $7 pot, especially if they want to use such a rare size!
Man there is quite a bit more time involved with building this than I originally expected!
My mje15xxxs are about 120 and 190 which is pretty good for mje15xxxs and as most HB builders used mjes these should work but I also have 2sa1930/c5171 which almost match perfectly - N and P about 140-150 so I've got a problem - which pair should I use?
So marlinx76 if you want almost a perfect pair go for 2sa1930/c5171. I bought mine some time ago from Mouser. They were made in Malaysia. A friend of mine uses these with 22pF in his amps and they do not oscillate so going up to 30pF just in case should be OK.
cheers,
So marlinx76 if you want almost a perfect pair go for 2sa1930/c5171. I bought mine some time ago from Mouser. They were made in Malaysia. A friend of mine uses these with 22pF in his amps and they do not oscillate so going up to 30pF just in case should be OK.
cheers,
I think your 3503 tested with 0.1mA base current will not be that far out.
If the hFE is 100 then the emitter current will be 1mA and for 200hFE ~2mAIe
these are fair Ie settings for a 100mA device.
the bigger problem is that the Tj will be very different for the two very different hFE. That extra variable destroys the usefulness of the DMM hFE facility.
Use a jig that gets the same Ie and the same Vce and the same Tc when you measure the Ib and Vbe.
0.35A continuous DC current requires a transformer with an AC rating > 0.7Aac.
This would run hot. Inside a ClassA amplifier (or hot ClassAB amplifier) that has a high Ta, it would probably run too hot.
Increase the transformer AC current rating to >1.4Aac and the transformer should run cool enough for long term reliability.
This would run hot. Inside a ClassA amplifier (or hot ClassAB amplifier) that has a high Ta, it would probably run too hot.
Increase the transformer AC current rating to >1.4Aac and the transformer should run cool enough for long term reliability.
The transformer is rated at 800VA (Antec 8445).
Don't get me wrong, I may be being a bit anal about this.
The hum emitted is only apparent within very close range.
The transformer is not getting hot at all, even after a couple of hours.
I am concerned about the overall consumption of the two amps in
their idle state, with view to leaving them powered. The reasoning behind
thr question is the fact they take at least 45 mins to stabalize after switch on
regarding the DC offset (20mV down to less than 1mV).
Thanks for your interest.
Don't get me wrong, I may be being a bit anal about this.
The hum emitted is only apparent within very close range.
The transformer is not getting hot at all, even after a couple of hours.
I am concerned about the overall consumption of the two amps in
their idle state, with view to leaving them powered. The reasoning behind
thr question is the fact they take at least 45 mins to stabalize after switch on
regarding the DC offset (20mV down to less than 1mV).
Thanks for your interest.
Don't leave the amp/transformer on. Problem with offset has nothing to do with the transformer. There must be something else. Once offset is adjusted on operational amplifier under normal working temperature of transistors it should stay close to zero with minimum variation.
If you have oscillocope check if there are no HF oscillations. Starting with cascode, it sometimes does such things and offset is firstly contrlled in LTP. There may also be some other reason for DC seeping into the output.
cheers,
If you have oscillocope check if there are no HF oscillations. Starting with cascode, it sometimes does such things and offset is firstly contrlled in LTP. There may also be some other reason for DC seeping into the output.
cheers,
That has to be your line ... or a slight DC on the ground of your input.
The DC blocker cap creates a "virtual ground" for your NFB node.
Whatever mV offset you have ... is the mismatch of your IP pair.
Of course , this can be nullified by the trimming of Re (offset adj.)
All "blameless" (cringe) topology's should stabilize to (whatever) mV
offset almost instantaneously.
OS
The DC blocker cap creates a "virtual ground" for your NFB node.
Whatever mV offset you have ... is the mismatch of your IP pair.
Of course , this can be nullified by the trimming of Re (offset adj.)
All "blameless" (cringe) topology's should stabilize to (whatever) mV
offset almost instantaneously.
OS