I wasnt very happy with the ccs, so made my own. it uses a 3v zener and the spare unused opamp channel that I had to control a bd139. Gives nice steady current. Please let me know how it looks. circuit is attached below.
Can I use a better reference, it needs to be more than 3v as ne5532 needs that much swing from the rails. I also have to look at the bd139 and see if its suitable for this. It might have to handle some 2-3w of power. Maybe the 3055 will be better for this.
I also tried a bjt mje3055 instead of irf510. Without feedback, from thd perspective, it works better than the mosfet for 32 ohm loads, but worse for 600 ohm loads. Will stick to irf for the time being, will test it out later and see if i like one over the other.
Can I use a better reference, it needs to be more than 3v as ne5532 needs that much swing from the rails. I also have to look at the bd139 and see if its suitable for this. It might have to handle some 2-3w of power. Maybe the 3055 will be better for this.
I also tried a bjt mje3055 instead of irf510. Without feedback, from thd perspective, it works better than the mosfet for 32 ohm loads, but worse for 600 ohm loads. Will stick to irf for the time being, will test it out later and see if i like one over the other.
Attachments
Possible issues...
The first stage will have a small but significant offset which is a problem as you have the volume control directly coupled to it (so could be noisy/crackly.) That DC is passed to all other stages although its less of a problem here because all the outputs are AC coupled.
(LTSpice doesn't show real world offsets with the basic models used. You could well find a couple of hundred millivolts offset depending on gain and volume setting)
The first stage will have a small but significant offset which is a problem as you have the volume control directly coupled to it (so could be noisy/crackly.) That DC is passed to all other stages although its less of a problem here because all the outputs are AC coupled.
(LTSpice doesn't show real world offsets with the basic models used. You could well find a couple of hundred millivolts offset depending on gain and volume setting)
LTSpice gives me some dc offset. I see about -45mv after the first stage. Same at mosfet out. When second stage also has gain, it gives about -36mv of offset. Not sure what to do with this but either put a cap or use opa2134. Input impedance is currently 47k, how low can I go safely without compromising much on input devices that I can connect to. This impact only when the volume is being changed using the pot and the noise is proportional to the dc offset, right?
Pots tend to go noisy when the wiper is passing DC current. Reducing the input impedance isn't really the answer imo (you could go lower, for example a modern CD player should easily drive at least down to 2k but that is not the way to go about it). Ideally an input impedance should be fairly high, say over 200k.
The easy answer is just to AC couple it and/or use FET device.
The easy answer is just to AC couple it and/or use FET device.
This link Coupling Capacitors in section 2 says that the pots make a noise when they are operated, I take that it means it wont add noise while listening to music, but only when I change the volume, i will hear noise.
In the sims, with a 1k pot, there is -1uA dc offset current in the wiper at 50% volume, goes upto -2.4uA at full volume. With a 10k, pot I can cut it down by a factor of 10 and hopefully that can cut the noise. lets see how much noise is there in a real build, then if needed, i will switch to 2134.
In the sims, with a 1k pot, there is -1uA dc offset current in the wiper at 50% volume, goes upto -2.4uA at full volume. With a 10k, pot I can cut it down by a factor of 10 and hopefully that can cut the noise. lets see how much noise is there in a real build, then if needed, i will switch to 2134.
I find the noise(a crackling or leaves rustling sound) most noticeable when making a large change in the volume, certainly not while it's in one position.
I have been playing around with the layout, attached is one. Not sure about the grounding yet. Should I create a separate audio ground, like I've done for headphones. I have not yet done the ground pour on the second layer, will do it in the end as the then they layers arent visible clearly. Please let me know how to improve it.
Attachments
You have labeled everything as GND.
Most of those "nodes" are actually returns for different signals.
Associate Every FLOW route with it's OWN RETURN route.
Keep that PAIR close coupled throughout their route across the PCB.
Do this for EVERY GND node.
Layout your PCB with these route defined PAIRs.
Do not use a GND pour with multiple slots as a ground plane.
Most of those "nodes" are actually returns for different signals.
Associate Every FLOW route with it's OWN RETURN route.
Keep that PAIR close coupled throughout their route across the PCB.
Do this for EVERY GND node.
Layout your PCB with these route defined PAIRs.
Do not use a GND pour with multiple slots as a ground plane.
What is a shield ground?
A coax is an unbalanced interconnect that passes the Flow part of the Signal through the core and passes the Return part of the Signal through the screen.
The "shield" as you call it is actually the Signal Return. It is NOT a GROUND.
If you are dealing with balanced signals in a 2core+screen, then the Flow (hot) and Return (cold) are on the two cores.
The screen which is connected to Pin1 of the XLR is connected to Chassis. Pin1 is NOT GROUND.
A coax is an unbalanced interconnect that passes the Flow part of the Signal through the core and passes the Return part of the Signal through the screen.
The "shield" as you call it is actually the Signal Return. It is NOT a GROUND.
If you are dealing with balanced signals in a 2core+screen, then the Flow (hot) and Return (cold) are on the two cores.
The screen which is connected to Pin1 of the XLR is connected to Chassis. Pin1 is NOT GROUND.
Look at C14.
The trace from input terminal to cap + is separated from the trace input terminal to cap -
That separation creates a LOOP.
Loop Area acts as an aerial that emits and/or receives interference.
I said:
But you can do better. You have gone for a two layer PCB.
You can place the Flow trace over the Return trace. This reduces the gap between the Flow and Return to the thickness of the PCB.
Do this analysis and minimise LOOP AREA for EVERY Flow and Return PAIR.
The trace from input terminal to cap + is separated from the trace input terminal to cap -
That separation creates a LOOP.
Loop Area acts as an aerial that emits and/or receives interference.
I said:
you can reduce the Loop Area by placing the Flow trace next to the Return trace.
But you can do better. You have gone for a two layer PCB.
You can place the Flow trace over the Return trace. This reduces the gap between the Flow and Return to the thickness of the PCB.
Do this analysis and minimise LOOP AREA for EVERY Flow and Return PAIR.
I did this for the headphone section, havent done for preamp yet. Please take a look and let me know if I am on the right track. I could not do much for the headphone output stage, the areas (with thick tracks) are still large due to physical size of heatsink, caps etc.
Edit: Ah, just figured out how to make o/p stage better.
Edit: Ah, just figured out how to make o/p stage better.
Attachments
Last edited:
Redid the layout. Images are attached. I have attached separate layer images for clarity. I still have some more work to do on the preamp.
I have a few questions:
1. I am currently using 3v zener to bais the ccs, any better option in this voltage range.
2. the opamp supply path is a bit long, so are the opamp outputs to transistors, is this an issue.
3. hp output return path is also long, is that an issue.
4. Is R20 the return path for R10?
Please let me know your comments.
I have a few questions:
1. I am currently using 3v zener to bais the ccs, any better option in this voltage range.
2. the opamp supply path is a bit long, so are the opamp outputs to transistors, is this an issue.
3. hp output return path is also long, is that an issue.
4. Is R20 the return path for R10?
Please let me know your comments.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.