Post by sadface on Nov 24, 2019 22:33:41 GMT 12
Hi Guys,
Here's a fun side project I've been playing with.
Due to my old backup preamp being gutted for use in another project (see the boombox thread), I have found the need to make a new backup preamp. I find I need a secondary preamp to throw into the system during my frequent urges for upgrades and modifications to my tube preamp.
also a KC5418 regulated supply which links don't exist for anymore.
My first attempt used the suggested tape equalisation as that sounded like the closest to a digital source..... thus I learned what tape equalisation actually was due to the horrible sound that came out of my speakers. I'm guessing the kit was designed a long long time ago as it has no suggestions for how to make a preamp without equalisation; the only options were tape, microphone or RIAA.
Here is what they suggested:
Here is what I first implemented: Note the 100k resistor after the 50k volume pot which cause some odd tracking.
Some jaycar parts were replaced with better salvage bits I had on hand. This caused bass rolloff issues as the 0.47uF and 0.33uF orange drops were a bit small in value.
After some research into how to get rid of the equalisation and set the gain appropriately I came up with this:
Note the simulated log pot with the 12k resistor after the 100k linear pot. sound-au.com/project01.htm Some brief testing proved this approach was superior to a cheap log pot.
Thus with a few parts omissions, resistors in capacitor holes and capacitors resistor holes I managed to make the board do what I wanted it to do rather than what it was designed for. I came up with this:
There was also a funny incident where I reversed the left channel wiring on the volume control which took me longer to diagnose than it rightfully should have.
I then replaced the orange drops with some Wima 1uF MKP caps which fixed the rolled off bass and I swapped out the LM833 for a NE5532 which sharpened up the sound quite nicely.
I can happily say that for the less than $100 worth of mats in here (salvaged box etc) the sound is unreasonably good. I'd would guess it would give some $500-$1000 preamps a bit of a run for their money.
|
Post by sadface on Nov 24, 2019 23:12:53 GMT 12
And now for the Mark II
With the hilarious success of the jaycar board I decided to up the game and make my own board to do exactly what I want.
The goal this time was to use both sides of a NE5532 for a single channel with the volume control nested in between the 2 stages.
Here is my schematic:
Overall gain should be about 2.1 iirc. I setup the jaycar board for a gain of 2 and it was about right with my power amps so I am sticking with the same here.
One reason for a mono board was that I am generally limited in horizontal space inside an amp enclosure but not limited vertically so the idea is to stack small footprint boards rather than use up more floor space for a stereo board. I would also like the option of a dual mono implementation.
So without further ado: may I present the mono preamp board. Its 99% done unless somebody with more knowledge than myself can see any flaws.
Decoupling caps are as close as possible to the IC.
Space has been allowed for large sized 5mm pitch film caps up to 10mm wide.
There is provision for dual output coupling caps for paralleling to make larger values.
All pads are as large as practical for ease of building.
Power entry is from the left side as this suits my current layout with the PSU on the left side of the enclosure.
Input from the right side suits the in/out jacks layout.
Output on the back for the shortest wiring out of the amp.
Volume pot connections are on the front for the shortest wiring to the volume pot (Mark II will use a shaft extension for the same purpose).
Just for a teaser, here is the power supply:
I also have a speaker protection circuit I am using to deal with turn off thump. The pcb is 1/2 done.
|
Post by sadface on Dec 13, 2019 21:56:17 GMT 12
Well Folks,
Some progress has been made.
I changed my mind on the feedback loop gain settings. I read that I will likely lose about 3db in the volume control and I've found with the current mark 1 I sometimes want a little bit more gain.
R4 and R8 have been change to 3k.
Gain should go from 2.15 to 3.00.
This means I'm now waiting on some replacement resistors which should have arrived today but will de delayed until Monday according to DHL.
Thankfully the parts for the protection board have arrived in the interim. It's a bit late for testing now but here is the completed board. The idea here is mainly to protect against turn on/off noises.
And here are the mono preamp boards waiting for the last resistors before the larger parts can be soldered.
I do love the look of Vishay/Dale RN series resistors. All resistors have been matched between channels as close as my old Fluke 25 can manage.
|
Post by colinf on Dec 14, 2019 6:53:48 GMT 12
Looks like a quality board. Just nitpicking, R2 should be 1.5k so that the impedances on the input pins of bipolar input opamp NE5532 are the same, to get lowest distortion. But it’s not really critical as the second stage has unmatched impedances anyway due to the varying output impedance of the volume pot. Also these circuits are quite prone to RF which can give them a reputation of being sonically unforgiving or even harsh. I usually use a small RF filter at the input. To get lowest distortion with this type of circuit a jfet input opamp could be used, like OPA1642 (with an adapter board) or even better, more costly types. Use an ic socket so you can swap opamps if need be. Even the lowly and ancient TL072 (dual version of TL071) can give reasonable sound. The studio up the road here has loads of them in the signal path and still manages to sound ok! Nakamichi used them in their cassette decks as well. Using a jfet input opamp you can eliminate the coupling caps as well, as no dc offset is developed as it is with a bipolar input opamp.
AMR-iFi R&D
|
Post by sadface on Dec 14, 2019 7:23:46 GMT 12
Hi Colin, Thanks for the feedback.
How would you implement an RC filter at the input? I do plan to use IC sockets as I plan to use LME49720 chips once the boards are tested and proven. I purchased a single LME49720 for use in the mark 1, however I decided not to use it as I would then have a burnt in chip in one channel with a brand new one in the other. I'll burn in both chips at the same time.
|
Post by sadface on Dec 14, 2019 10:30:19 GMT 12
The first test has been passed
The relays click off almost immediately upon power down.
Strangely: they don't have an audible click on. However I've tested with a multimeter for continuity and the output is definitely disconnected when un-powered so I can only assume that this is a quirk of the particular relays.
Hopefully I will have time after this afternoon to test the dc detection. Saturday drudgery comes first.
|
Post by colinf on Dec 14, 2019 19:49:11 GMT 12
You can add an RF filter by just putting in a 270pf cap after R2 (on pin 3 of the IC) going to ground. That would roll off the HF response at about 587kHz, assuming a source impedance of zero ohms. You can reduce it as source impedance increases. The LME49720 is a bipolar input opamp but it has a unique input bias cancelling circuit, so that it doesn’t generate dc offset where the NE5532 does. So you can get away with eliminating the coupling caps with that opamp, similar to jfet input opamps. You can even force the output stage to operate in class A by adding a constant current source diode (CCS) going from the output pin (1 and 7 for the two stages) to negative (or positive) power. The CCS could be set at say 5mA which would allow class A up to output of around 7v rms with the 2.2k feedback resistor.
AMR-iFi R&D
|
Post by sadface on Dec 21, 2019 8:34:00 GMT 12
Here's the completed preamp pcbs awaiting testing.
Unfortunately I had a mishap last night.
I soldered a ZTX951 PNP transistor in the spot for a ZTX851 NPN.
In the process of trying to get the bugger back out again I lifted 2 pads. Now I am truly sadfaced.
Lucky for me I have another 4 boards and plenty of spare parts so all I really lost was time and self esteem.
|
Post by colinf on Dec 21, 2019 20:47:10 GMT 12
To get transistors out of plated through holes it’s likely best to heat all 3 pins at the same time and wriggle it out. To heat all 3 pins use a shaped bit of solid copper wire to transfer heat from the soldering iron tip to the other 2 pins at the same time. That way you shouldn’t overheat the tracks as it’s relatively quick.
AMR-iFi R&D
|
Post by sadface on Dec 23, 2019 11:54:14 GMT 12
Well folks,
The power supply is built and ready for testing.
Now I just need to rig up a jig for testing and break in.
|
Post by sadface on Jan 13, 2020 21:08:21 GMT 12
G'day guys,
IT VERKS!
The speaker protection circuit is now tested and works perfectly.
Turn on delay is about 5 seconds.
I've tested the detector by applying DC from power supply outputs through a 100k resistor into the detector. The relays are tripped immediately upon contact with either supply rail and through both the left and right channel detector inputs.
The loss of AC detector trips the relays immediately upon power off.
I will call this a roaring success.
Tomorrow I will hookup the preamp boards for testing and see if it all makes appropriate noises.
I've decided I will use the existing boards to start off with. However I will build another pair of boards and omit the coupling capacitors to see how it changes the sound.
Colin, given that it the LME49720 has dc blocking on the input, does it then follow that I can omit the coupling caps on the output also?
|
Post by sadface on Jan 13, 2020 21:14:15 GMT 12
I also discovered that the inrush current on a Jaycar 15VAC 20VA transformer must be at least 150mA.
Since I could only get quick blow fuses in a 3AG from Jaycar this morning, I purchased 100mA, 150mA, 250mA and 500mA fuses, a couple of each just in case.
The 100mA and 150mA fuses blew immediately upon power up but the 250mA fuse has held firm through testing so far.
Once the preamp goes into the proper chassis, I'll go back to the 100mA slow blow M205 fuse used in the current Mark I version.
|
Post by sadface on Jan 13, 2020 21:34:01 GMT 12
Also I seem to have fixed the issue where the relays don't audibly click on.
Maths error was the cause. I need to drop 3v from the 15v supply rail into the 12v relays. I worked off the current draw for one relay but the pair is wired in parallel so I was using double the resistance required and dropping 6v instead of 3v.
G5V-1 relay = 12vdc coil @ 960ohm I = 12v/960ohm =0.0125A
2x 0.0125A = 0.025A Therefore R=3v/0.025A = 120R.
P = e2/r = 9/120 = 0.075W or 75mW
I previously used a 240R 2W resistor I did things the lazy way and tacked a 2nd 240R resistor onto the first
I don't think I will have any power dissipation issues with 2x 2w resistors in parallel passing 75mW of current.
With the extra 3v on the coil, they now click on audibly.
|
Post by colinf on Jan 13, 2020 23:59:33 GMT 12
You could use a 3.3v zener diode in place of the 120 ohm resistor if you like as it drops 3.3v whether the relays are on or not. The LME49720 has a bipolar transistor (bjt) input stage, like the LM833 and NE5532, that draw input bias current and thus generate dc offset in the signal path, that needs to be blocked by coupling capacitors. But the LME49720 (and its twin LM4562) is unique in that it also includes small constant current sources on the input stage that cancel the bjt input bias current, unlike the other 2 opamps. So you can get away without coupling caps for the LME49720 only, not the LM833 or NE5532. Jfet input opamps (like TL072, OPA2134, OPA1642) also don’t need any coupling caps that can colour the sound. But be aware the circuit now amplifies dc so if your source has dc offset on it, it will be amplified by 3 (according to your gain settings) at full volume.
AMR-iFi R&D
|
Post by sadface on Jan 14, 2020 21:39:39 GMT 12
Well lads,
It makes noise!
I'll give it a couple of evenings of run time to burn things in a bit then I'll move the test jig into my main system and see if the sound is up to scratch.
From what I hear behind me out of the boombox I think detail is going to be a strong point.
|
Post by sadface on Jan 14, 2020 21:48:15 GMT 12
I should ask.
Is it necessary for it to actually drive a load for burn in purposes or could I simply leave it playing with no power amp attached?
|
Post by Owen Y on Jan 15, 2020 8:10:10 GMT 12
When testing, I plug in say a 47k to 100k R across ea output +/- (wire up a couple of RCA plugs). It may not be essential for this preamp, but it simulates an amp load.
|
Post by colinf on Jan 15, 2020 11:18:17 GMT 12
Solid state stabilises over time. It’s got to do with bipolar transistors’ electron migration effects inside the silicon die which gradually even out, and lower distortion. So you could leave it on without a load and it will still stabilise. Once the silicon gets into a similar pattern of thermal expansion and contraction over a couple of power-up cycles it can be considered run in. I’ve done experiments with this though and music running through it at low volume seems to work better than no music. It definitely seems to work better after you switch the power on and off a few times over the day. The LME49- series like to be left on, sounding steely when cold.
AMR-iFi R&D
|
Post by sadface on Jan 16, 2020 20:43:12 GMT 12
Well boys,
The Backup Preamp just made its way into my main system for a proper audition.
Shucks but it sounds. All sorts of good stuff going on in the mid-range. Bass response resonates the house as required. Highs are clear and by no means brittle which I often find is an issue with solid state gear.
The LME49720 is seemingly a bloody good opamp.
I could be the chip, it could be the circuit, it could also be the stupid low noise power supply. I'm going to guess that it is a bit of each.
Either way this sounds a lot nicer than $100 worth of materials.
Now that my work bench no longer has the boombox on it I can start building a set of preamp boards minus the coupling caps to see what it does to the sound.
I do have some spare boards if anybody wants to have a play.
|
Post by sadface on Jan 16, 2020 20:48:06 GMT 12
The use of a proper Alps pot might also have made a significant difference compared to the jaycar fake log pot setup from the mark I
|
Post by colinf on Jan 16, 2020 22:15:00 GMT 12
Volume pot quality does make a big difference. Leave the LME49720 powered up overnight then listen to it, see what you think.
AMR-iFi R&D
|
Post by sadface on Jan 17, 2020 15:59:38 GMT 12
It also occurred to me that I should build another version without coupling caps but using carbon comp resistors instead of metal film to how it sounds.
It's quite a luxury having multiple sets of boards to experiment with different components.
|
Post by sadface on Jan 18, 2020 12:36:27 GMT 12
Has anybody ever used Takman Carbon Film Resistors?
Do they justify their extra cost over something like say KOA Speer 2% carbon film resistors?
|
Post by Owen Y on Jan 18, 2020 14:28:45 GMT 12
I have some, but not tried yet, sorry. I have used Takman Met films & they are fine, no complaints. Generally I do not use costly components anymore. Good audio quality, yes, but no Vishay naked foils or Duelands - rarely does a device justify such disproportionate expense (IMHO). The cct design comes first.
|
Post by colinf on Jan 18, 2020 15:42:39 GMT 12
I used Takman carbons in my own stuff, they sound ok, more depth and far quieter than generic carbons, and the resistance tolerance is excellent. They have a distinctive sound, see if you like it or not. I like it, but too many of them in the signal path can be a bit overbearing. The Takman metalfilms have a similar character but not quite as strongly.
AMR-iFi R&D
|
Post by colinf on Jan 18, 2020 15:50:13 GMT 12
Owen, Duelunds are worth it! But I agree they could be less costly. The best sound is no cap though, and if you can design the circuit without caps you make it both better and less costly. But sometimes you just can’t beat a good cap and a simple circuit. More elaborate circuits can be good though, just more difficult to implement.
AMR-iFi R&D
|
Post by sadface on Jan 19, 2020 14:26:00 GMT 12
G'day Chaps,
Mark III is now up and running. The LME49720 chips were recycled from the Mark II so no burn in required.
No coupling caps I added a Low Pass RC filter on the input. I used a 330pF C0G ceramic cap that I had in stock soldered across R2 and the nearest ground. Not the prettiest, but I have a feeling I will order a new set of boards with additional features by the time I am done.
I work this out as 482kHz corner frequency.
I probably should have implemented the no caps and RC filter separately to see their individual effects.
The main difference in the sound is noticeable decrease in steely harshness at the top end. Exactly what I was hoping for.
The next step will be to order some carbon resistors and see what that does to the sound.
"You can even force the output stage to operate in class A by adding a constant current source diode (CCS) going from the output pin (1 and 7 for the two stages) to negative (or positive) power. The CCS could be set at say 5mA which would allow class A up to output of around 7v rms with the 2.2k feedback resistor."
Would a E-562 diode be suitable? This appears to be the cheapest CCS diode available around 5mA.
Does it matter which supply rail the diode is connected to? I am guessing this only effects the orientation of the diode?
I struggle a bit with diode polarity. Whilst I know how to wire them up correctly, their actual polarity seems to be the reverse. That is to say, the band is would be wired to the positive supply rail in a bridge rectifier however this is the negative end on a spec sheet.
|
Post by colinf on Jan 19, 2020 19:18:56 GMT 12
What source are you using? The source’s output impedance changes the RF filter frequency. Try listening without the 330pf cap, see if you can hear the difference, it should be worse. RF affects the sound of opamps detrimentally. 330pf is quite a lot, I generally use around 100pf. You can substitute your Clone Note for the volume pot in your circuit if you like. An E-562 constant current diode would work fine. You can connect it one of 2 ways. 1. With the anode connected to the opamp output pin and the cathode (end marked with a band) connected to the - power rail (pin 4 on the opamp). Or 2. With the cathode connected to the opamp output pin and anode connected to + power rail (pin 8 on the opamp). Note that a constant current diode isn’t a conventional diode, it’s a JFET with selected maximum drain current (called Idss). For the E-562 that’s 5.6mA. They just call it a diode because it has two connections. Remember a rectifier diode passes current one way. The electron output side of a diode is the cathode, marked with a band. The schematic symbol for a diode shows which way the current flows with a solid arrow. The line next to the arrow tip (the cathode) shows that current can’t flow from this end against the arrow. The voltage across a diode when current is flowing through it (said to be forward biased) is about 0.6v. When the diode is reverse biased no current flows and the voltage across it can be anything up to the rated reverse voltage of the particular diode. A 1N4004 diode has a reverse voltage rating of anything up to 400V, and a forward current rating of 1 amp. Schottky diodes have been designed to have less forward bias voltage across them, about 0.2 to 0.3v, so they dissipate less power for the same current as a normal diode and can be made smaller.
AMR-iFi R&D
|
Post by sadface on Jan 22, 2020 20:51:34 GMT 12
Hi Colin,
My source is a Musical Fidelity X-Dac V3.
"Audio output at digital 0dB 2.2VRMS at 1kHz nominal, 50 Ohms impedance" from the spec sheet.
Would the 50ohm output impedance be considered in series with R2 (1k) to make 1050ohm or am I missing something here?
How long could the corner frequency be set without effecting operation for audio purposes? My thinking is that even set at say 200kHz I am not filtering anything important to audio.
I'm not sure what you mean by a Clone Note.
I'm plotting a new version of the pcb with an RC filter properly implemented on the input and space for E-562 diodes.
|
Post by colinf on Jan 23, 2020 9:58:41 GMT 12
Yes, it’s in series with the input resistor to make 1050 ohms. I usually go not less than 10 times the frequency you want, in this case 20kHz. But you can make it less, it just starts to make the top end softer. There’s less and less negative feedback available for correcting the signal at frequencies above the audio band in most audio opamps. So the less HF you put through one the better the sound. A Clone Note is a passive volume control made up of a light dependent resistor in series with the signal and a normal resistor of a certain value switched to ground to set the volume. It’s meant to sound better than a normal volume pot.
AMR-iFi R&D
|