I have a few amps that I have been playing with for a while, then reediting, selecting correct transistors, learning new things and then updating the schematic. In my IPS my LTP is made from Linear's JFET LSK489. While I get the idea of what transistors are a good fit and where, I however didn't really drill down into the small details. I would like this amp to be fast and clean - the rest of the transistors in the critical are robust enough to allow an A bias as well as an AB bias without issues. I am far from an electronics expert but I would imagine that the IPS is possibly the most critical stage of an amplifier - I chose a JFET because of the how of input impedances - the LSK489 is also very quite and is able to handle more voltage then the 389 while having less capacitance making it is easier to prevent from oscillating. Because I would like this to be a high speed design I felt that both a constant current source and sink with transistors would be the best way to do it....what I haven't read up on is what the best devices are to function as CCS/S in the IPS and what parameters are most beneficial however in my schematic I elected to choose bipolar transistors instead of diodes however I am using a zener to protect the JFET from too much voltage. Over the years I have seen just about anything in this area from BC devices to MPSA devices and so on. Other then my voltage requirements, what parameters are clutch for this area and critical to performance? What would allow the IPS to be as quick and accurate as possible?
thanks in advance,
Jared
thanks in advance,
Jared
What information would help? I imagine the IPS portion is the most critical piece of the design - I would like it to be as quiet as possible while being as fast as possible. I am guessing that both a CCsource and sink would be the best way to make sure that is is sourcing and sinking as much current as fast as possible while seeing a stable supply of current without anything restricting it. I plan to use bipolar devices for this part - my VAS is a combo of bipolar devices and mosfets, my driver and OPS is separated to use higher voltages on the IPS and VAS. Driver is a MOSFET and the outputs are LFETS.
Goldmuld used MSPA42 for this portion which would is nice because they are robust and handle higher voltages (I got 1000s) if and when I design a larger version of this. The first one I make will be within LSKs voltage requirements to test it out see if I like the sound and make sure it actually works. What are better options for these transistors then MPSA42 for the CCSource/sink or is that pretty much as good as it gets?
Now, if you consider the output section is basically a couple of transistors across the power supply rails.......do you still think the IPS is the most critical piece of the design?
Neither does he. Forum gained tech education flops again.
Book reading a must. Self, Cordell, Pass articles.
I am trying to be polite and have been for years, because I know I get penalised if I am too honest.
Book, schmook. Back in my day we didn't have books. Self's and Cordell's hadn't been written. The only "theory" you could find started with a load line, then went into six pages of differential equations describing the operation of a two transistor push pull thansformer coupled stage - in way more detail than anyone ever cared to go. No computer anyone could afford would run the real Spice - and that student version which ran off two floppy disks would crap out at nine transistors. Models? If you wanted one you kludged it from a data sheet.
So how did you learn to build amplifiers? By building amplifiers. You still do. But it never hurts to read the books we have nowadays.
The input stage IS the most critical. If the voltage/current differencing does not work right everything goes to hell quickly and no advanced VAS or output stage will save you. But it's also not THAT hard to get it right - and if you read the books and build a few amplifiers you'd figure that out for yourself.
End rant. Every amp you build doesn't have to be perfect. Just the last one. I've built plenty that never went beyond a pcb with wires hanging out of it and a transformer sitting on a bench. The good ones get finished.
Proposed schematics do facilitate discussions.
What is your source impedance driving the input?
Jfets tend to be used mostly on very high source impedance like magnetic phono or tape head. For CD player, radio, computer output, cell phone output, they are overkill IMHO. An extra stage.
What is your bandwidth? for DIY audio it tends to be 20 khz or 40 khz to be total nyquist compliant with the best of baby's ears.
Don't assume the worst on junction transistors. I bought some run of mill ON semi 2n5401, find they work great with a 1kohm collector load with 270k pull up & 400 k pull down on base. That is I think a 100k input impedance (which I needed to be driven by tube anode of a PAS2). Datasheet says these are 50 gain parts. So try your MPSA42's & 92's I'm saying, unless you're building radio receivers. You might be surprised.
What is your source impedance driving the input?
Jfets tend to be used mostly on very high source impedance like magnetic phono or tape head. For CD player, radio, computer output, cell phone output, they are overkill IMHO. An extra stage.
What is your bandwidth? for DIY audio it tends to be 20 khz or 40 khz to be total nyquist compliant with the best of baby's ears.
Don't assume the worst on junction transistors. I bought some run of mill ON semi 2n5401, find they work great with a 1kohm collector load with 270k pull up & 400 k pull down on base. That is I think a 100k input impedance (which I needed to be driven by tube anode of a PAS2). Datasheet says these are 50 gain parts. So try your MPSA42's & 92's I'm saying, unless you're building radio receivers. You might be surprised.
The reason you have seen anything in this area from BC to MPSA is that device choice here is not critical to performance if the circuit is competently designed and you use decent small signal transistors of the types commonly used in audio, such as KSC1845/KSA992 (what I use), 2N5551/2N5401 (favoured by Cordell as shown e.g. in his LSK489 app note, which you may find useful, see fig. 11) and many others.
The only thing I can think of is Self's finding that the MPSA42 isn't happy operating at too low a Vce such as the current mirror LTP load of the Blameless topology, although I don't know (and iirc he doesn't mention) if the same applies to the MPSA92 if your CM is at the top, as I imagine it will be with your JFET input. Other than this, stick to the usual suspects and you'll be fine.
Cheers,
Cabirio
Yes, I do. Regardless of how good any of the other areas of the amp are they will never be able to make up for a degraded and bad input signal. You cant polish a turd.
How is my question faulty? I am asking what are the best devices for a CCS/S in the IPS portion of an amp and if the IPS having both a source and sink that is transistor based is the best way to make use of the LSK489 I am using in the LTP. What makes you feel that I haven't read anything on the topic? Pass is why I chose a JFET and Cordell is why I looked at the LSK series. I know enough to not use the 389 because of the capacitance of it and don't want to get into cascoding or fixing it so does that allow me to ask questions on a forum? I read a ton on speaker design but I couldn't voice a good speaker until after building a few and seeing what works and doesn't. I'm sure the same applies to amps and pretty much everything else - regardless if it is electronics or woods and metals. I never made any claims that I am an expert on the topic and prefaced it upfront asking a question backed with reasoning to get a better understanding - the correct response would be recommending an appropriate device or disagreeing and stating why, possibly recommending something else unless of course you don't have the answer. So here's your opportunity to shine and show off your engineering prowess.
I can post one. I want the amp to be able to handle any device the same so I picked a JFET. I would like the amp to be able to produce infrasonic low end so it can work for home theater applications as well as music reproduction. 40khz is fine on the high end but would prefer to have a bandwidth of 3hz to 100khz. To be totally honest I wanted to experiment with the bandwidth as a way to replace active crossovers but I need to get a circuit that performs well before I play with that area. The only thing that i don't like about using a JFET is that it needs to be protected when using it with high voltage rails. It may not be needed but I would like the amp to perform the same regardless of the previous components. I do a lot of pro audio stuff and use my turntables daily but the mixer acts like a pre for it so I guess a JFET isn't really needed for it. I was looking at a bunch of different transistors and 2SA1163 and 2SC2713 look like they would be a good one for a number of different areas in a amp. If I change the input transistors I would do something that is fully balanced and symmetrical. I don't know enough about the topic to pick an input impedance and any number I throw out would be arbitrary, I planned on laying out the topology first and then simming the circuit with a bunch of different values to see what works best - there's a designer that I really admire and enjoy his products - he uses a 40k input impedance for balanced connections so I would probably start there and play with it.
Open loop BW and pole of IPS ought to be far beyond 100KHz although the input low pass filter rolloff won't be that high. Differential input would greatly reduce common mode noise. For SS, Jfet is my choice. THe high input impedance of Jfet allows use of polypropylene or similar quality input caps to be used, typically about 1uf. Mr Cordell covers a few good examples in his book. I would keep Vds of the Jfet less than 15V, 10V. (Cascode) how the IPS associates with the rest of the sections within the closed loop is quite important in determining functionability . Type of frequency compensation (COG/NPO) caps for example. The PCB traces and layout is also part of the circuit that requires extra detail in simulation to include, but a poor layout with regard to frequency compensation can be a disaster for sure and leave you with bald patches atop your scalp.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.