Hi Ronny,
thanks for explanation! Do you remember where you first saw the circuit? Or which amplifiers use similar ones?
Yes, most similar amplifiers don't use active loads, but simple resistors instead and also don't use buffers in between LTP and VAS (LTP).
I don't find the buffers that important, but a buffer with common mode control of the active load in the LTP seems useful.
I like to keep the CCS for the second (VAS) LTP as is and your suggestions inspired me to this circuit instead:
My circuit idea follows the same thinking like the circuits you presented. Not sure whether mine works however. I need to study all circuit options in more detail.
Indeed, next PCB is revision #4 and I hope to get all right this time. There will be changes in the area we are discussing, I'm just not sure how exactly.
thanks for explanation! Do you remember where you first saw the circuit? Or which amplifiers use similar ones?
Yes, most similar amplifiers don't use active loads, but simple resistors instead and also don't use buffers in between LTP and VAS (LTP).
I don't find the buffers that important, but a buffer with common mode control of the active load in the LTP seems useful.
I like to keep the CCS for the second (VAS) LTP as is and your suggestions inspired me to this circuit instead:
My circuit idea follows the same thinking like the circuits you presented. Not sure whether mine works however. I need to study all circuit options in more detail.
Indeed, next PCB is revision #4 and I hope to get all right this time. There will be changes in the area we are discussing, I'm just not sure how exactly.
Nice, and plenty hedrom for the next Vas stage The emitter resistors should flow 10 + * 10mA. for full "modulation".
8mA first stage, to me its high, and the buffer transistors may as well flow somthing imbetween its imbetweeeen ? i dont know if its nessesery youre pretty close to vas current.
Of topic:
First time i saw it it was not in a magasine it was a print out. And since i have been active radioamateur it was a lot of tecnical stuff and magasines that made som attention. At that time us hifi tecnical students even borrowed hifi to dismantle it, making drawing, a pensil scetch of the pcb, after the weekend we delivered it back "sounds good" but not to expectation... 😉
The first amplifier i made was with 4 psc 572 tubes in paralell for 3-30 Mhz Radioamateur bands 600Watt output.
Sorry my typing thats part of my handycap.
8mA first stage, to me its high, and the buffer transistors may as well flow somthing imbetween its imbetweeeen ? i dont know if its nessesery youre pretty close to vas current.
Of topic:
First time i saw it it was not in a magasine it was a print out. And since i have been active radioamateur it was a lot of tecnical stuff and magasines that made som attention. At that time us hifi tecnical students even borrowed hifi to dismantle it, making drawing, a pensil scetch of the pcb, after the weekend we delivered it back "sounds good" but not to expectation... 😉
The first amplifier i made was with 4 psc 572 tubes in paralell for 3-30 Mhz Radioamateur bands 600Watt output.
Sorry my typing thats part of my handycap.
Thanks for your support!
The low Vce of the transistors is a real issue and fixing this greatly improved the performance. All it takes is one resistor.
I set up a small test jig (see attachment) to play with different options.
What I found so far is that all options, means the original Cordell circuit, the modified Cordell circuit with "R7" to increase the Vce and also the circuit I came up with based on your suggestions behave well in the AC domain. Mine has ~6dB lower gain, but this is something I don't mind.
Here is the AC response:
Green = original Cordell
Blue = Cordell modified with extra resistor to increase Vce
Red = My Frankenbuffer
Petrol = standard current mirror
THD is also interesting:
With 1mV input:
Normal current mirror: 0.7%
Cordell original: 0.6%
Cordell modified: 0.02% (30-fold improvement!)
My Frankenbuffer: 0.002%
With 2mV input:
Normal current mirror: 1.5%
Cordell original: 14%
Cordell modified: 0.1%
My Frankenbuffer: 0.008%
Seems like the original Cordell circuit was over-driven the way I set it up. May be related to me, not the circuit. The extra resistor helps here, too.
THD is not really an apples to apples comparison since my circuit outputs only half the voltage swing, thus THD should be doubled to be fair.
THD is not everything, spectrum is also important. This is the spectrum with 1mV input:
I have no idea why my circuit is so good. Actually I don't even really understand it.
I tried to increase gain of my circuit, but this seems not a good idea. Either this causes weird looking roll-off or a peak, both not looking good. Better live with lower gain and enjoy low THD. I wonder whether there is some feedback loop somehow or what else causes an order of magnitude lower THD.
Regarding the bias of the buffer: I don't know what is a good trade-off here. I just chose 5mA per transistor as a middle ground. I can't imagine that the VAS draws that much (nonlinear) current to justify a buffer at all.
The tail current is chose at 8mA so that I can apply plenty of LTP emitter degeneration, extending linear range and increasing slew rate. I could lower the current and decrease degeneration, but why?
Trying to understand what is going on, I have redrawn the circuit another way:
I still don't get it. Does anybody recognize this circuit somehow?
The low Vce of the transistors is a real issue and fixing this greatly improved the performance. All it takes is one resistor.
I set up a small test jig (see attachment) to play with different options.
What I found so far is that all options, means the original Cordell circuit, the modified Cordell circuit with "R7" to increase the Vce and also the circuit I came up with based on your suggestions behave well in the AC domain. Mine has ~6dB lower gain, but this is something I don't mind.
Here is the AC response:
Green = original Cordell
Blue = Cordell modified with extra resistor to increase Vce
Red = My Frankenbuffer
Petrol = standard current mirror
THD is also interesting:
With 1mV input:
Normal current mirror: 0.7%
Cordell original: 0.6%
Cordell modified: 0.02% (30-fold improvement!)
My Frankenbuffer: 0.002%
With 2mV input:
Normal current mirror: 1.5%
Cordell original: 14%
Cordell modified: 0.1%
My Frankenbuffer: 0.008%
Seems like the original Cordell circuit was over-driven the way I set it up. May be related to me, not the circuit. The extra resistor helps here, too.
THD is not really an apples to apples comparison since my circuit outputs only half the voltage swing, thus THD should be doubled to be fair.
THD is not everything, spectrum is also important. This is the spectrum with 1mV input:
I have no idea why my circuit is so good. Actually I don't even really understand it.
I tried to increase gain of my circuit, but this seems not a good idea. Either this causes weird looking roll-off or a peak, both not looking good. Better live with lower gain and enjoy low THD. I wonder whether there is some feedback loop somehow or what else causes an order of magnitude lower THD.
Regarding the bias of the buffer: I don't know what is a good trade-off here. I just chose 5mA per transistor as a middle ground. I can't imagine that the VAS draws that much (nonlinear) current to justify a buffer at all.
The tail current is chose at 8mA so that I can apply plenty of LTP emitter degeneration, extending linear range and increasing slew rate. I could lower the current and decrease degeneration, but why?
Trying to understand what is going on, I have redrawn the circuit another way:
I still don't get it. Does anybody recognize this circuit somehow?
Looks a bit similar, doesn't it?
Snip from a set of slides titled "EE4345 transistor current sources and active loads"
It looks like Cordell's CM distortion reducing circuit, doesn't it?
I posted in another thread started by @bimo where he uses this circuit (the thread is mistakenly named 'HEC'), see link below.
I also suggested to fix the resistor divider point voltage so as to disable the CMR circuit and find out what difference it makes so one can learn from what it does and not does, it was just an idea suggested on a whim and needs to be thought out how to make it work, if possible at all, op suggested it wouldn't work or maybe there was some misunderstanding in communicating the idea.
https://www.diyaudio.com/community/threads/simple-hec-amplifier-24vdc.388896/post-7087218
Last edited:
R67 and R71 can be seen as a mirror, R68 and R71 can be seen as a mirror.
Now having two mirrors in paralell common current feedback is related to R67 and R68 in paralell 50 Ohm and R71 at 100.
Would have thought R67 // R68 = R71 makes a common current feedback ratio, also bias at 2nd stage makes U R71 equal Ubase emitter in the common current feedback or? Was there som latch up mentioned? Like a bistable "mirror" ? Then R67 // R68 should not be lower than R71 ? Or ?
Now having two mirrors in paralell common current feedback is related to R67 and R68 in paralell 50 Ohm and R71 at 100.
Would have thought R67 // R68 = R71 makes a common current feedback ratio, also bias at 2nd stage makes U R71 equal Ubase emitter in the common current feedback or? Was there som latch up mentioned? Like a bistable "mirror" ? Then R67 // R68 should not be lower than R71 ? Or ?
Last edited:
like u see it Requires a smal current gain from first to 2.d stage.
Also i belive your current flow in the front end can cause current noise, so i wuld keep it outside the front end as much as possible.
I stil like somthing close to the original.
Also i belive your current flow in the front end can cause current noise, so i wuld keep it outside the front end as much as possible.
I stil like somthing close to the original.
Playing around with those circuits was fun and interesting, but once a reality check is done, it pretty much falls apart.
I tried substituting a random transistor for the BC557C instead of BC557B and here it shows that it no longer works as expected.
The Cordell circuit is most sensitive to mismatch. It may even fail to work properly at all. The book authors quoted earlier wrote that matched transistors would be best for the Cordell circuit and simulation confirms it. This one seems to behave worst with transistors not matched.
The Cordell circuit with the extra resistor is way more tolerant towards mismatch. The extra resistor helps here, too. Great!
The new circuit is also pretty sensitive to mismatch. All magic low THD performance is gone with slightest mismatch.
Well, this happens. Something seems great in simulation, but does no longer work with real life tolerances.
Good thing is that all circuits are just assembly variants. By adding or removing resistors, three variants can be built.
Planning in matched transistors would be nice, but such parts are extremely expensive and hard to find.
SMT dual transistors may work well, even ones that are not matched since they should be from the same wafer.
Regarding the CMRR of the different load options, I made some plots.
Both inputs are shorted together and driven by an AC voltage source.
See for yourself:
Surprising observations are:
All active loads are no better than resistor loads. No idea why.
The simple current mirror performs worst even with perfectly matched components.
The emitter follower enhanced CM is great in theory, but also fails in reality.
The simple resistor load with 1% tolerance preforms pretty well.
I'll make up my mind about dual THT and SMT transistor options and also how to possibly improve real world performance with unavoidable component tolerances.
I tried substituting a random transistor for the BC557C instead of BC557B and here it shows that it no longer works as expected.
The Cordell circuit is most sensitive to mismatch. It may even fail to work properly at all. The book authors quoted earlier wrote that matched transistors would be best for the Cordell circuit and simulation confirms it. This one seems to behave worst with transistors not matched.
The Cordell circuit with the extra resistor is way more tolerant towards mismatch. The extra resistor helps here, too. Great!
The new circuit is also pretty sensitive to mismatch. All magic low THD performance is gone with slightest mismatch.
Well, this happens. Something seems great in simulation, but does no longer work with real life tolerances.
Good thing is that all circuits are just assembly variants. By adding or removing resistors, three variants can be built.
Planning in matched transistors would be nice, but such parts are extremely expensive and hard to find.
SMT dual transistors may work well, even ones that are not matched since they should be from the same wafer.
Regarding the CMRR of the different load options, I made some plots.
Both inputs are shorted together and driven by an AC voltage source.
See for yourself:
Surprising observations are:
All active loads are no better than resistor loads. No idea why.
The simple current mirror performs worst even with perfectly matched components.
The emitter follower enhanced CM is great in theory, but also fails in reality.
The simple resistor load with 1% tolerance preforms pretty well.
I'll make up my mind about dual THT and SMT transistor options and also how to possibly improve real world performance with unavoidable component tolerances.
BTW in your cordell you had 2n5551 and bc5* as npn and complentry swap! FOR input trans. It couldnot preform.
Also like your 200 & 100 Ohm resistor's for input they wolud simplify and dont latch due to smal "internal emitter resistors". Thats also the reason it preforms! 200 Ohms paralelled with 200 Ohms = 100, choose bias current to be equal to Ube= 0.65V R = 100, I zero = 6.5mA, for input idle. stil high, 150and300 Ohm would fit somthing with differentcurrent settings.
Also like your 200 & 100 Ohm resistor's for input they wolud simplify and dont latch due to smal "internal emitter resistors". Thats also the reason it preforms! 200 Ohms paralelled with 200 Ohms = 100, choose bias current to be equal to Ube= 0.65V R = 100, I zero = 6.5mA, for input idle. stil high, 150and300 Ohm would fit somthing with differentcurrent settings.
Attachments
You're right there was a (deliberate) mistake in the simulation: I swapped one of the transistors for another type to see the impact of mismatch.
My simulation files are a terrible mess, sorry for that.
As I just learned (again), I should randomly swap transistor models more often to check how tolerant the circuit is towards mismatch. Back then I forgot to swap the transistor type back to BC5XX.
Regarding the bias current I would say it is more or less a matter of taste. Also whether to use BJT or JFET in the LTP. So far I have not observed issues with noise even when listening to the idle amplifier using sensitive headphones. But that was a few revisions ago.
The project is about exploring the maximum performance that can be obtained based on this topology.
It can be stripped down and made much simpler; simpler than Cordell's design (which is excellent BTW).
Most DIYers prefer simple designs.
I do the opposite: Add features, complicate things, risk stability and so on. And I often failed with the added complications.
Meanwhile I learned that it is pointless to just observe the whole amplifier when adding or changing something. It is better to take it apart and study and optimize each part on its own and then put it back together and see how it interfaces with the other circuitry and how it reacts with each other and optimize again.
Meanwhile I decided to add SMT transistors in parallel to small signal THT transistors as cost optimized option for (more or less) matched transistors. Being in the same package, this helps with thermal stability, too. And it makes the design more versatile, which I always prefer. Usually I try to plan in fallback options to simpler methods just in case I prefer that.
My simulation files are a terrible mess, sorry for that.
As I just learned (again), I should randomly swap transistor models more often to check how tolerant the circuit is towards mismatch. Back then I forgot to swap the transistor type back to BC5XX.
Regarding the bias current I would say it is more or less a matter of taste. Also whether to use BJT or JFET in the LTP. So far I have not observed issues with noise even when listening to the idle amplifier using sensitive headphones. But that was a few revisions ago.
The project is about exploring the maximum performance that can be obtained based on this topology.
It can be stripped down and made much simpler; simpler than Cordell's design (which is excellent BTW).
Most DIYers prefer simple designs.
I do the opposite: Add features, complicate things, risk stability and so on. And I often failed with the added complications.
Meanwhile I learned that it is pointless to just observe the whole amplifier when adding or changing something. It is better to take it apart and study and optimize each part on its own and then put it back together and see how it interfaces with the other circuitry and how it reacts with each other and optimize again.
Meanwhile I decided to add SMT transistors in parallel to small signal THT transistors as cost optimized option for (more or less) matched transistors. Being in the same package, this helps with thermal stability, too. And it makes the design more versatile, which I always prefer. Usually I try to plan in fallback options to simpler methods just in case I prefer that.
Wel... here i end up with a caskomp having infinite base emitter resistors like current sourses. and that is huge current gain, equal to the transistors used. Then it doesnt even look close to the original !?!?!?!?
Dumping the buffers standing current (which is variable since the load current is subtracted from the standing current) into the tail node does not make this a cascomp. According to a quick simulation, it seems this improves gain, but also worsens THD and PSRR.
I thought about a cascomp LTP some time ago. The cascomp is very clever using feed forward correction and thus very linear, but may introduce stability issues I would not know how to deal with. Instead, I settled for CFPs, which work by negative feedback, may also go unstable, but I would know how to deal with instability. The negative feedback loop of the CFP work well into pretty high frequency range (unlike the global feedback of the amplifier) and thus lowers distortion well beyond the audible range. The CFPs reduce THD by an order of magnitude roughly and performance doesn't depend on matching.
Besides preparing the PCB for SMT dual transistors as assembly option, I'm still trying to figure out how to improve the load / buffer circuitry to perform better with mismatched components. I would be willing to add another two transistors for that.
I thought about a cascomp LTP some time ago. The cascomp is very clever using feed forward correction and thus very linear, but may introduce stability issues I would not know how to deal with. Instead, I settled for CFPs, which work by negative feedback, may also go unstable, but I would know how to deal with instability. The negative feedback loop of the CFP work well into pretty high frequency range (unlike the global feedback of the amplifier) and thus lowers distortion well beyond the audible range. The CFPs reduce THD by an order of magnitude roughly and performance doesn't depend on matching.
Besides preparing the PCB for SMT dual transistors as assembly option, I'm still trying to figure out how to improve the load / buffer circuitry to perform better with mismatched components. I would be willing to add another two transistors for that.
Have you tryed linearise inout stage with 5 10 Ohms each side?
You can do part of the vas load by using mos drive.
Or though ab your favor the cascomp predive ?
You can do part of the vas load by using mos drive.
Or though ab your favor the cascomp predive ?
Good point to avoid (non-linear) loading of the VAS.
I heard that MOSFET (pre-) drivers are great from a lot of experienced designers so this may be a good option in deed. I plan to experiment with that some day. Important would be to bootstrap the transistors collectrors driven by the VAS to avoid Vgd dependent capacitance loading, which is signal swing dependent. Thus, a bootstrapped diamond buffer could be an option. I could build such a diamond buffer on the PCBs I have either with BJTs or MOSFETs. I failed with bootstrapping the diamond buffer earlier, but meanwhile figured out how it should work. A bootstrapped diamond buffer made of BJTs also presents a constant capacitance to the VAS, but draws more current.
I can't find smaller HEXFETs than the IRF610 / 9610. I find them a bit too heavy for a pre-driver, but good as drivers. This makes a MOSFET diamond buffer a bit unattractive.
Further refinements would be the bootstrapped VAS clamps by Glen Kleinschmid. I tried to make this work well, but failed. However, this made me reconsider the floating cascode that I added to the VAS because this draws a lot of current from the VAS output. This was a brain fart and I grounded the cascode now, but creating a virtual ground for that.
Over the weekend I made up a lot of weird active load circuits. The idea was to fix the issues with open loop DC operating conditions and poor CMRR, which was partially successful. I found some funny solutions. The DC issues seems to be an issue when running the circuits open loop, which also affects open loop THD performance, but the issues mostly go away once the global loop enforces proper DC conditions. Conclusion is that the "new" active load is overall a good trade off with fall-back options to the (improved) Cordell load and plain resistor load. In terms of THD and CMRR, no active load beats plain resistor load, especially with mismatched transistors. But the active loads are able to generate a lot of gain, which is not reasonably feasible with a resistor load. I'll spare you the details unless you are really interested.
The idea to downgrade to a plain resistor load is tempting. I had a look at how the whole amplifier would perform with the LTP loaded by plain resistors.
I made a comparison to see the difference. The variant with resistor load is a copy of the latest schematic with active load so everything else is equal.
Here is the loopgain (TPC, 700kHz UGLF). Active load on the left, resistors on the right:
No surprise that OLG is much lower, over 30dB lower. At 1kHz, there is ~20dB less OLG available for lowering THD. The amplifier now has a wider open loop gain, which is sometimes seen as something desirable (with little scientific reasoning behind AFAIK).
Note I intentionally mismatched all transistors in the front end circuit that would be good to match. Global feedback takes care of the resulting DC imbalance. Other performance metrics will be affected however.
Simulated DC offset with active load is -35mV (was something below 10mV with matched transistors) and -3mV with LTP resistor load. Thus NFB mostly irons out the DC issues.
Here is a THD comparison between LTP load options (TPC, 700kHz UGLF) with 75Vpp into 4R load:
THD 1kHz resistor load: 0.00011%
THD 1kHz active load: 0.00011%
THD 20kHz resistor load: 0.0022%
THD 20kHz active load: 0.0022%
This is surprising because I would expect that 20dB of OLG do make a difference at 1kHz THD.
Maybe this is because the active load adds some distortion. So there is higher OLG, but more inherent distortion with the active load? Or 70dB OLG is just enough and 90dB therefore yields no improvement?
However having a lot of gain in the lower frequency range does not seem to be something really useful to me.
I really consider to actually build the next variant with resistors as load. Why add complexity if it does not pay off?
I heard that MOSFET (pre-) drivers are great from a lot of experienced designers so this may be a good option in deed. I plan to experiment with that some day. Important would be to bootstrap the transistors collectrors driven by the VAS to avoid Vgd dependent capacitance loading, which is signal swing dependent. Thus, a bootstrapped diamond buffer could be an option. I could build such a diamond buffer on the PCBs I have either with BJTs or MOSFETs. I failed with bootstrapping the diamond buffer earlier, but meanwhile figured out how it should work. A bootstrapped diamond buffer made of BJTs also presents a constant capacitance to the VAS, but draws more current.
I can't find smaller HEXFETs than the IRF610 / 9610. I find them a bit too heavy for a pre-driver, but good as drivers. This makes a MOSFET diamond buffer a bit unattractive.
Further refinements would be the bootstrapped VAS clamps by Glen Kleinschmid. I tried to make this work well, but failed. However, this made me reconsider the floating cascode that I added to the VAS because this draws a lot of current from the VAS output. This was a brain fart and I grounded the cascode now, but creating a virtual ground for that.
Over the weekend I made up a lot of weird active load circuits. The idea was to fix the issues with open loop DC operating conditions and poor CMRR, which was partially successful. I found some funny solutions. The DC issues seems to be an issue when running the circuits open loop, which also affects open loop THD performance, but the issues mostly go away once the global loop enforces proper DC conditions. Conclusion is that the "new" active load is overall a good trade off with fall-back options to the (improved) Cordell load and plain resistor load. In terms of THD and CMRR, no active load beats plain resistor load, especially with mismatched transistors. But the active loads are able to generate a lot of gain, which is not reasonably feasible with a resistor load. I'll spare you the details unless you are really interested.
The idea to downgrade to a plain resistor load is tempting. I had a look at how the whole amplifier would perform with the LTP loaded by plain resistors.
I made a comparison to see the difference. The variant with resistor load is a copy of the latest schematic with active load so everything else is equal.
Here is the loopgain (TPC, 700kHz UGLF). Active load on the left, resistors on the right:
No surprise that OLG is much lower, over 30dB lower. At 1kHz, there is ~20dB less OLG available for lowering THD. The amplifier now has a wider open loop gain, which is sometimes seen as something desirable (with little scientific reasoning behind AFAIK).
Note I intentionally mismatched all transistors in the front end circuit that would be good to match. Global feedback takes care of the resulting DC imbalance. Other performance metrics will be affected however.
Simulated DC offset with active load is -35mV (was something below 10mV with matched transistors) and -3mV with LTP resistor load. Thus NFB mostly irons out the DC issues.
Here is a THD comparison between LTP load options (TPC, 700kHz UGLF) with 75Vpp into 4R load:
THD 1kHz resistor load: 0.00011%
THD 1kHz active load: 0.00011%
THD 20kHz resistor load: 0.0022%
THD 20kHz active load: 0.0022%
This is surprising because I would expect that 20dB of OLG do make a difference at 1kHz THD.
Maybe this is because the active load adds some distortion. So there is higher OLG, but more inherent distortion with the active load? Or 70dB OLG is just enough and 90dB therefore yields no improvement?
However having a lot of gain in the lower frequency range does not seem to be something really useful to me.
I really consider to actually build the next variant with resistors as load. Why add complexity if it does not pay off?
I upgraded the npn differential pair in my schematic to cfp and it gives a clear improvement in thd starting from 10W output power.
My amp topology is totally different, could be same, better or worse in another amp.
Added a second pnp per side over the cfp, the same way as the pnp over the npn and it is even a tiy bit better.
Only sim, not build.
My amp topology is totally different, could be same, better or worse in another amp.
Added a second pnp per side over the cfp, the same way as the pnp over the npn and it is even a tiy bit better.
Only sim, not build.
Way to much phase swing.
So if nohing happens with 20 dB more loop gain, the unlinnear pice is the input.
So if nohing happens with 20 dB more loop gain, the unlinnear pice is the input.
Clever idea to use lateral MOSFETs since they tend to stabilize their bias at a certain current.
However, they present a massive (gate to drain) capacitance to the VAS output.
Bootstrapping makes this capacitance constant, but also maximizes it.
Good that I build this amplifier is a modular way so I can easily experiment with different arrangements.
I spend some time simplifying some parts of the latest circuit, updating the PCB design accordingly and cleaning up the PCB which was cluttered and partially congested due to adding and removing circuitry countless times. I also simulated a lot of scenarios and meanwhile I'm somewhat confident that everything is optimized (Well, I was at this point three times already...). I'm going to order new PCBs prior to end of the year.
The current schematic allows building a lot of different variants of this front end. I'm confident that one of them is going to work well.
At this point I like to thank everybody who helped me with this project:
@Ketje
@chalky
@AKSA
@tiefbassuebertr
@triplej
@nowatt
@r_merola
@nattawa
@OldDIY
@drinkingcube
@hpasternack
@bimo
@Bernhard
@levinson mark
@R Dijk
@RCruz
@peufeu
...and many more forum members I forgot to mention.
Thank for inspiration. Thanks for getting my head straight whenever I was stuck.
The end (of the year) is near and this is a good opportunity to reflect the past year. While I made a lot of progress and feel like on the homestretch, this is still a challenging project and without you I probably would not have made it that far.
The current schematic allows building a lot of different variants of this front end. I'm confident that one of them is going to work well.
At this point I like to thank everybody who helped me with this project:
@Ketje
@chalky
@AKSA
@tiefbassuebertr
@triplej
@nowatt
@r_merola
@nattawa
@OldDIY
@drinkingcube
@hpasternack
@bimo
@Bernhard
@levinson mark
@R Dijk
@RCruz
@peufeu
...and many more forum members I forgot to mention.
Thank for inspiration. Thanks for getting my head straight whenever I was stuck.
The end (of the year) is near and this is a good opportunity to reflect the past year. While I made a lot of progress and feel like on the homestretch, this is still a challenging project and without you I probably would not have made it that far.