Hi Alan,
Oops! You're absolutely right about that. Thanks.
Hi astx,
As for signals up around 350,000Hz, it doesn't take much of a stray capacitance at higher impedance's to reduce the level way down. Were you using a X10 probe when looking at the output, or was is connected directly to the 1 Mohm input? If you were simply directly connected, that level might be a lot higher than you think it is.
-Chris
Oops! You're absolutely right about that. Thanks.
Hi astx,
As for signals up around 350,000Hz, it doesn't take much of a stray capacitance at higher impedance's to reduce the level way down. Were you using a X10 probe when looking at the output, or was is connected directly to the 1 Mohm input? If you were simply directly connected, that level might be a lot higher than you think it is.
-Chris
On Mouser today I spied new P channel Jfet:
InterFET SMD/SMT P-Channel JFETs JFET | Mouser
Datasheet says June 2019.
InterFET SMD/SMT P-Channel JFETs JFET | Mouser
Datasheet says June 2019.
It is apparent from the spectrum analyzer that above a certain freq, the 'stuff' comes thru again untouched.
THx-RNMarsh
Last edited:
DAC output HF filtering
Dear Chris,
the output of the pcm5122 DAC is low ohmic and followed by a simple output filter (470R resistor and 2.2nF). See datasheet excerpt.
A standard oscilloscope probe (200MHz, 1X set) was connected from the output filter RCA side to the 1MOhm input of the spectrum analyzer.
(so the "stray" capacitance of the cables adds to the 2.2nF)
Amplifier input impedance is about 40kOhm. I think we can trust the measured levels.
BR, Toni
Dear Chris,
the output of the pcm5122 DAC is low ohmic and followed by a simple output filter (470R resistor and 2.2nF). See datasheet excerpt.
A standard oscilloscope probe (200MHz, 1X set) was connected from the output filter RCA side to the 1MOhm input of the spectrum analyzer.
(so the "stray" capacitance of the cables adds to the 2.2nF)
Amplifier input impedance is about 40kOhm. I think we can trust the measured levels.
BR, Toni
Attachments
Hi Toni,
That's a lot better than what I thought it was going to be. Your output impedance is therefore going to be about 500 ohms or so. Not too bad.
Being the careful tech I am, just for yourself, you should measure the output voltage at some of the higher frequencies to prove to yourself that it's flat. I've had some nasty surprises in the past when equipment performs differently than the specs would suggest. It never hurts to actually check and make certain the claims are true.
So, how do you like the sound of that one?
-Chris
That's a lot better than what I thought it was going to be. Your output impedance is therefore going to be about 500 ohms or so. Not too bad.
Being the careful tech I am, just for yourself, you should measure the output voltage at some of the higher frequencies to prove to yourself that it's flat. I've had some nasty surprises in the past when equipment performs differently than the specs would suggest. It never hurts to actually check and make certain the claims are true.
So, how do you like the sound of that one?
-Chris
There was a radical approach to the 'stuff' that appeared in The Audio Amateur 3/94 issue by Kalman Rubinson - "Passive Filters For Digital Audio" using coils for a DAC project.
Due to the cost of a DAC and shielding issues in a CD player this didn't get to fly with me however I have kept all of my vinyl and CD collections and have some interest in getting rid of the 'stuff'.
The approach Kalman adopted was neither for a Butterworth nor a Bessel pattern but something in between.
The graphs in the article represented SPICE simulations with Miller, Renco and Toko coils. The main illustration used Miller 9250 coils and I have shown the form of this in a dummy .asc file
Some references were Analogue Devices Application Note AN-327; Burr-Brown Application AB-026A; Ultra-Analogue Devices Application Note AP-02
... beside the datasheet values about THD+N I think this combo "raspberry + pcm5122" (e.g. hifiberry, justboom ...) offers an outstanding performance for the low price.
I'm using it as "party machine" where the master volume level is controlled by a rotary switch and the software volume as well as the player can be controlled remote using MPDroid or M.A.L.P. accessing the mpd server on the raspberry. An nginx web server delivers "folder.jpg" files to the 2 different clients.
BR, Toni
Last edited:
Hi rsavas,
Apologies for the late reply - thanks for pointer. The changes should not be too bad for any input bias currents if i use low value resistors (comparatively), and if both resistors have the same ppm temp coeff, it should be somewhat mitigated.
Regards,
Shadders.
The use of larger-wattage metal film resistors in the feedback network is to reduce distortion, particularly distortion at low frequencies caused by signal-dependent temperature changes in the resistors. Having the same tempco in the feedback resistor and the feedback network shunt resistor does not necessarily help, since they do not see the same amount of power dissipation (unless extraordinary measures are taken with a network of numerous resistors).
This may be guilding the lilly, but the performance of the feedback resistors in a power amplifier is very important, and it is wise to be on the very conservative side.
Cheers,
Bob
Hi Bob,
Thanks for the reply - i ran the simulation - and for the values used - the resistor power ratings are 2x that required - so the footprint on the PCB allows for higher value resistors as per your book, and the power dissipation will be significantly lower.
As an aside - i used surface mount where possible, and this is proving to be a pain to fault find if something goes wrong. The FMBM5401 is a dual transistor package - 2x 2N5401 transistors - and a SOT package.
They have stopped selling the TO-92 package, and only Farnell sell them as their own brand Multicomp. Finding TO-92 packages with the 150volts Vceo is difficult in the UK.
Regards,
Shadders.
Hello Shadders,
I have been using 2n5401 from Rectron. Mouser sells them as 583-2N5401-F
Central semi also makes them, costing a bit more, Mouser sells them too, as 610-2N5401. Both are well stocked right now.
Can you explain why fault finding is more difficult using smt? or is it a case of replacing smt is harder than TH equivalents?
Rick
I have been using 2n5401 from Rectron. Mouser sells them as 583-2N5401-F
Central semi also makes them, costing a bit more, Mouser sells them too, as 610-2N5401. Both are well stocked right now.
Can you explain why fault finding is more difficult using smt? or is it a case of replacing smt is harder than TH equivalents?
Rick
Hi rsavas,
Thanks for the guidance about Mouser components.
I used SMT as it provided a smaller board space - i am building a 3-way active speaker and 6 amps (class A/B) in the same enclosure - reducing space required as far as possible.
If i use Mouser - have done so many times - it is shipped from the US - free, even for $40 ?? worth of goods. Funnily, even though it is shipped from the US to UK via air, the total cost is cheaper than Farnell UK. Stock is better for mouser too.
Yes - replacing an SMT device will destroy the old - no issue, but getting to the device if there are other components around it such as capacitors, is difficult. Even fault finding - a DMM probe can be a bit tricky.
What i have done is build the bare bones of the amplifier - so access to the parts is easier - to test different values - as i have found out, that simulation may not be as accurate as i expected - even for DC.
Regards,
Shadders.
Thanks for the guidance about Mouser components.
I used SMT as it provided a smaller board space - i am building a 3-way active speaker and 6 amps (class A/B) in the same enclosure - reducing space required as far as possible.
If i use Mouser - have done so many times - it is shipped from the US - free, even for $40 ?? worth of goods. Funnily, even though it is shipped from the US to UK via air, the total cost is cheaper than Farnell UK. Stock is better for mouser too.
Yes - replacing an SMT device will destroy the old - no issue, but getting to the device if there are other components around it such as capacitors, is difficult. Even fault finding - a DMM probe can be a bit tricky.
What i have done is build the bare bones of the amplifier - so access to the parts is easier - to test different values - as i have found out, that simulation may not be as accurate as i expected - even for DC.
Regards,
Shadders.
I guess that depends on the technique. I am able to replace most smt without damaging them. Need to control the heat, be fast and use good equipment or technique. I do have a hot air machine however, but only use that for application where an iron will not work. I do use two irons for smt passives, make solder bridges, using solid copper wire forms for others.
What design are you using for your amps? BC-1 is a fine amp, I think appropriate for your 6 amp app.
Yes, use of SMT parts for DIY projects is a double-edged sword. They save a lot of space, but can be difficult to handle and troubleshoot. For those of us who are older ("more experienced
, when going with SMT resistors we can use the larger ones like 0805 or 1206 and still save a ton of room. Importantly, if using SMT resistors, especially in the signal path, make sure you get metal film ones, not the ordinary thick film ones. Never hesitate to use a through-hole part in amongst mostly SMT parts if that part is critical or likely subject to change.
Cheers,
Bob
Hi Rick,
You are lucky to have the hot air machine -i have a Weller soldering station - basic one only. The SMT parts are cheap - so destroying is not too much a problem, as i always over order parts too.
The amplifier design is my own - based on the basic amplifier in Bob's book - i added a cascode for the VAS. I have two boards working - but when i added the second set of output transistors, the amplifier oscillated. So, i took the bare bones approach to only add those parts needed to test a working amp - as i anticipate my mitigation for oscillation when simulating, may also be the problem in the real world. I also wanted to change values to see what happens.
I assume the BC-1 is on this website ?. Will search for it. I have the boards for my design - so will progress these - as it is a learning experience.
Regards,
Shadders.
Hi Bob,Yes, use of SMT parts for DIY projects is a double-edged sword. They save a lot of space, but can be difficult to handle and troubleshoot. For those of us who are older ("more experienced
Importantly, if using SMT resistors, especially in the signal path, make sure you get metal film ones, not the ordinary thick film ones. Never hesitate to use a through-hole part in amongst mostly SMT parts if that part is critical or likely subject to change.
Cheers,
Bob
I have used 1206 in most areas - their power rating is much more than needed - so adds headroom if changes increase power dissipation.
As you have said - always use thin film - one of Doug Self's books showed the noise inherent in thick film - although thick film are much cheaper.
If i redesign the board - will use through hole components in specific areas - such as signal paths. I have used the rear of the board for current mirrors and sources, and the DC servo - saves a lot of space.
Regards,
Shadders.