Post by sadface on Sept 15, 2018 12:50:57 GMT 12
Hey Guys,
A nice easy build this time round.
The missus expressed a desire for some kind of small stereo device to run off an ipod in the bedroom. One could always buy a perfectly serviceable example of such a device for probably very little money on trademe. But that is no fun.
Off we go to jaycar to purchase 2x I then changed tack and ordered some over-sized psu filter boards off ebay with the intention of dual mono giant filter banks: 10000uf, 2200uf, 100uf,100nf per supply rail.
I then changed tack again and ordered some regulated psu pcbs that looked like I could refit them into a set of serious, regulated psus
A couple hundred dollars worth of parts from RS components and element 14 later, I started to look at the all the premium caps and the pair of transformers rather jealously. Then came the realisation that its all going to play mp3s off an ipod through some cheaps drivers I found in Jaycar. The dual regulated supplies and upgraded parts for the amp boards started to look a bit ridiculous in context and so all that stuff goes towards some fancy gainclones for myself in the not too distant future.
So i whacked together the pair of amp boards with their original parts kits.
A couple of complaints about these boards; the holes for mounting the lm3876 ics on the board are offset to far from the edge and necessitate some rebending of the pins to get the chip to line up with the edge of the board.
I also don't like the use of pins for I/O connections, I would prefer some solder on screw terminals. I also don't like the small gauge pcb holes in general, they don't allow for much modification of the design or upgrading to larger components.
Next time I will use LM3886 boards from China, these can be gotten for something like $5 for a pair including shipping on ebay. The chinese boards don't include a zobel network or the other network of a wire wrapped around resistor who's name i cannot remember (Thiel network?) but that is not the end of the world as these can be wired off board and these boards look a bit better laid out and appear to have heavier traces, fatter through holes and larger solder pads. I'll report back on this later when I get my hands on some.
Back to the original psu board I got from jaycar. Turns out it was pretty basic; a diode bridge and 2x 4700uf 50vdc caps was it. The rest of the board was dedicated to turning +/-40vdc into +/-15vdc.
No matter, a bit of head scratching later I figured out I could jerry-rig the board to make something suitably robust with some half decent filtering.
Jumper wires fitted up ready for solder.
I'm going to use the output at the 15VDC end but I will re purpose some of the holes that are supposed to house resistors to allow me to fit some extra 2200uf caps. 100nf caps have been soldered on the underside to allow clearance of the 2200uf caps on top of the board.
If i don't procrastinate too much this weekend I should have a working Mark 1 version in a crude MDF enclosure by the end of the weekend.
The finished product as a wee way off I think.
Firstly, it needs a proper stained wooden box to meet my aesthetic requirements. I've got some spare Okume veneer plywood from a desk I made earlier in the year which should solve that problem. Secondly, it needs a vacuum tube preamp stage. This is not technically required but it would be boring without one. I think I'm settled on one of these as it look hilariously simple 1 extra board, 1 extra power tranny and viola. Watch this space.
6DJ8s at USD2 each makes the tube cost negligible.
Also, a good looking source for NOS russian military caps for those in the know.
Thirdly, it needs a radio receiver circuit. Preferably AM/FM but FM only could suffice.
I have done some cursory searching on google and it seems like basic kits aren't expensive online. However, me being me. I want something along the lines of an 'audiophile' grade radio if such a thing exists. I'll admit I know next to nothing about radio circuitry at this stage so I will ask the question.
Is there such a thing as an 'audiophile' grade radio or is a radio generally about as good as another radio?
|
Post by Owen Y on Sept 15, 2018 14:28:31 GMT 12
Radio receivers are quite complex - perhaps pick up a decent SS FM tuner s/hand?
|
Post by michaelw on Sept 15, 2018 16:00:18 GMT 12
not all radios are the same ! there are definitely ordinary ones and audiophile ones and there are ordinary ones stuffed into a fancy box pretending to be audiophile (from companies who should know better !) 
|
Post by sadface on Oct 18, 2018 15:56:55 GMT 12
G'day Guys,
An overdue update.
Mark I of the Gainclone Boombox is up and running.
An oversized crude mdf box. Very pretty!
Here's a half decent long shot of the internals. Taken after initial testing of the modules.
|
Post by sadface on Oct 18, 2018 18:52:53 GMT 12
 Rectifiers, generic 400v/35A $3 jobbies.
My beautiful braided psu wiring.
Star ground.
A module semi mounted to one of the heatsinks. No insulating bush or mica washer or thermal grease.
An inside shot of the completed amp side.
Inside shot of the power supply side
 Inside wide shot. Notice the large empty space in the middle where a preamp stage might live in the near future.
Thus far it has been built to be taken back apart. Open back, extra long wire runs, plenty of room to move bits around.
Speaker drivers are cheap Jaycar jobbies
|
Post by sadface on Oct 18, 2018 19:20:53 GMT 12
It's been running since Saturday. It's mostly plays Radio Concert NZ off my cowan flac player while it gets some burn in hours.
My verdict so far, is that this sounds unreasonably good for what it cost to build.
It sounds best with classical music rather than modern stuff. It is a bit lacking in the bass department possibly due to the cheap drivers, possibly due to the poorly optimised cabinet.
The mid-range and highs however are downright lovely.
The next step is maybe an upgraded power supply, maybe a valve preamp. Eventually both.
My cowan plays radio sufficiently well I won't bother with a radio circuit at this stage.
It's essentially a testbed at this point in time.
Once I have nailed down the psu and preamp stages I will redesign the cabinet to fully encase the electrical bits and hopefully optimise the sound as much as can be.
I wouldn't mind downsizing the whole thing as well. Crude cabinet is crude and bulky.
Of course the redesigned cabinet will get some prettier looking wood with a nice slathering of Danish oil.
Maybe upgraded drivers too, I'll be keeping an eye on trademe for bargain full range drivers. I thought briefly about adding an outboard sub woofer but It just doesn't suit the application.
|
Post by Owen Y on Oct 18, 2018 20:10:42 GMT 12
Good work! Yes, maybe tune/port the box more optimally for reflex bass? Enough heatsinking for the B-Rectifiers?
|
Post by sadface on Oct 19, 2018 11:43:20 GMT 12
Hi Owen,
Plenty of heat sink on the rectifiers I think.
I’m not running it very loud so perhaps at high load they might get warmer but the metal brackets the are mounted on are cold to touch during regular operation.
|
Post by sadface on Oct 21, 2018 14:07:47 GMT 12
Hi Guys,
There appears to be some nice vintage drivers floating around on trademe right now.
Anybody know much about these?
I had an idea how they might be used in something like an infinite baffle arrangement.
Does this have any potential to sound good?
Possibly too sensitive to placement effects given its likely to get jammed into a corner?
|
Post by colinf on Oct 21, 2018 23:13:11 GMT 12
Hi sadface, interesting project. Is the Mrs into the details of construction as well? Just reading that you didn’t use heatsink paste when mounting the LM3876. It’s best to use just a little, even if there is no insulator. That way you won’t get hotspots on the back of the IC where the metal doesn’t quite reach the heatsink. The rectifiers won’t need any heatsink for a power output of this size. The heatsink with no insulator will be at V- potential. You can also get plastic case versions of the 3876 and 3886 which don’t need insulators.
AMR-iFi R&D
|
Pundit
Post by papahemi on Oct 22, 2018 7:08:52 GMT 12
Hi Guys,
There appears to be some nice vintage drivers floating around on trademe right now.
Anybody know much about these?
I had an idea how they might be used in something like an infinite baffle arrangement.
Does this have any potential to sound good?
Possibly too sensitive to placement effects given its likely to get jammed into a corner?
I have used the Richard Allan CG12S in open baffles, the baffle will need to be large to get acceptable bass response. The 12" also rolls of in the top a lot earlier than the 8" (CG8T are my favoured Richard Allan Driver) and I felt that they needed to have a tweeter. Mids are good tho'.
We are what we pretend to be, so we must be careful about what we pretend to be.
|
Pundit
Post by papahemi on Oct 22, 2018 11:43:17 GMT 12
 Richard Allan specs may be of interest
We are what we pretend to be, so we must be careful about what we pretend to be.
|
Post by sadface on Oct 22, 2018 15:05:44 GMT 12
Hi sadface, interesting project. Is the Mrs into the details of construction as well? Just reading that you didn’t use heatsink paste when mounting the LM3876. It’s best to use just a little, even if there is no insulator. That way you won’t get hotspots on the back of the IC where the metal doesn’t quite reach the heatsink. The rectifiers won’t need any heatsink for a power output of this size. The heatsink with no insulator will be at V- potential. You can also get plastic case versions of the 3876 and 3886 which don’t need insulators. Hi Colin,
That was simply a dry fit.
I used a standard insulating bush and mica washer arrangement.
I also cleaned the contact surfaces and the washer with an iso-prop swab, before giving them a nice gob of Arctic Silver 5 on both sides of the mica washer.
The missus will have a fair say in the final construction and aesthetic details.
|
Post by sadface on Oct 22, 2018 15:14:43 GMT 12
Hi Guys,
There appears to be some nice vintage drivers floating around on trademe right now.
Anybody know much about these?
I had an idea how they might be used in something like an infinite baffle arrangement.
Does this have any potential to sound good?
Possibly too sensitive to placement effects given its likely to get jammed into a corner?
I have used the Richard Allan CG12S in open baffles, the baffle will need to be large to get acceptable bass response. The 12" also rolls of in the top a lot earlier than the 8" (CG8T are my favoured Richard Allan Driver) and I felt that they needed to have a tweeter. Mids are good tho'. A pair of the CG8Ts came up on Trademe recently but got outside of my meager budget.
Thanks for the info on the 12s. They were probably a bit silly for the application. I'll keep an eye out for more 8's.
|
Post by colinf on Oct 23, 2018 4:18:29 GMT 12
|
Post by sadface on Oct 26, 2018 13:22:16 GMT 12
Another update.
I've been on holiday this week so I've had a bit of time to mess about. I've gone in circles a fair bit about what kind of tube stage to add but first I wanted to see how it might work out in the first place.
I made a spare set of half decent rca cables out of cat 5e cable and "premium" Jaycar connectors.
Proof of concept complete. No finesse involved this time round, maybe later.
Due to Jaycar closing and me running out of time, I ended up with the wrong sized tube core and the wrong sized braided sleeve. I had to skip a couple of steps in the process but the overall result was downright serviceable if not attractive.
All this allowed me to put my old backup tube preamp in front of the boombox.
As one would expect, the sound became immediately nicer on the ear; warm and rich like a fine aged rum. Since this is a low cost build and I am simply going after "pretty darn good" and "warm" rather than anything truly audiophile this preamp fits the build perfectly. The added bonus here is now I also get to build a new backup preamp.
With the preamp stage decided, its time to decided what I want to do about upgrading the power supply and how to put it all in a pretty box.
|
Post by sadface on Oct 26, 2018 17:32:38 GMT 12
Gentlemen, Behold!
The Guanzo XY Hifi 2.0! I got this a decade or so back when I was in uni. My very first tube amp!
All the finest materials I could afford around my uni-student booze budget.
A steel chassis from analogue metric with my first hackjob attempt at a wooden front panel. I never got around to doing it properly as I quickly began to make plans for a real valve preamp.
I paid about $160 for the assembled board,tubes and transformer on trademe, which was a bit of a rip off. It turns out that the same thing is still available on ebay for all of $80 including shipping.
I quickly got rid of the cruddy pcb mounted volume pot and replaced it with a stepped attenuator from the now deceased DIY Fidelity in Australia.This made a noticeable improvement.
I also quickly got some Nos tubes to replace the cheap chinese 6n3s that came with it. I got a pair of Ericson 2c51 which were a huge improvement. These later wore out and now I have some NOS RCA Command 5670s which are equally good.
When I plugged it into my system at the time everything just sounded better; all sparkly and shit. All of my uni mates thought the same. The mid-range just had something special.... Probably some good ol' distortion but who cares.
Bass is rolled off at the bottom. Every now and Mr Guanzo serves duty while my regular preamp is on the healing table for modifications. Whenever I plug the big preamp back in I feel like I've gained an extra octave at the bottom (M&K sub makes the difference very noticeable).
Top end is pretty nice, no complaints here, possibly mildly rolled off as when I think back to my old rotel preamp it seemed incredibly bright in comparison.
All and all not terrible attributes for a preamp that will now serve dedicated duty driving some full range drivers that are also rolled off at the bottom and probably the top and so won't noticed the absence of either.
|
Post by sadface on Oct 26, 2018 17:52:21 GMT 12
I have one concern with this preamp and that is the lack of any kind of soft start.
Attached is an alleged schematic for the psu. I will soon pull this all out and inspect the board to confirm before it goes on the healing bench. I will say it looks right from a cursory inspection of the board.
Pretty basic in all. Possibly serviceable.
Seems to me like the instant b+ from solid state rectifier is going to be pretty mean the cold tubes at startup. Possibly extended tube life to be gained for an improved B+ system. Filament supply looks grossly insufficient by modern standards. Needs at least some kind of voltage regulation maybe some extra caps.
Here is an alleged schematic of the signal stage. Once again I haven't confirmed yet but it looks about right.
Seems like quite an odd design. I'm not qualified to comment on anything other than the very odd volume pot location.
I might replace the yellow coupling caps with some Soviet PIO caps for shits and giggles but otherwise I don't plan to fiddle with the signal stage.
|
Post by colinf on Oct 27, 2018 7:32:43 GMT 12
You could always add a valve rectifier in series with the output side of the diode bridge for soft HT startup. Such as an EZ80. Put both anodes together in parallel. It would need an extra 600mA heater current. Or perhaps use a high voltage TO220 case mosfet (with 10v zener diode for gate protection) with a capacitor connected to ground and resistor to B+.
AMR-iFi R&D
|
Post by Owen Y on Oct 27, 2018 10:41:35 GMT 12
Vol pot is usually found at the front input of line stages, but an argument for putting it at the output is that S/Noise should be better without attenuation at the input. But of course, output impedance will vary with vol setting & will be quite high - if a a lower R pot is used (such as 15k) then it may work OK with your power amp. (I think I've got that right  )
|
Post by sadface on Oct 27, 2018 12:43:02 GMT 12
Hi guys,
These 2 issues are why I decided to upgrade to a better valve preamp rather than modify the existing design.
The soft start can bed fixed, I just need to decide on exactly how to implement it.
The volume pot location has always led me to lump this design into the fatally flawed basket for my main system. Over on diyaudio.com some people have tried moving the volume pot to the input side but it hasn't been successful.
I've considered changing to a smaller pot, the lm3875 spec sheet recommends a 10k log pot. The downsides to this would be downgrading from a quality stepped attenuator to a potentionmeter(no money in the budget here). Also I don't know how this is going to change the behaviour of the preamp.
It think its quite likely that half the lovely sound it makes is the tubes being driven outside of their linear range. It ran for many years as is without issues driving different SS amps so I think I'll be leaving this bit alone.
|
Post by sadface on Oct 27, 2018 14:25:52 GMT 12
So now we're back to the power supply side.
So far I'm looking at a couple of different directions. Option a) is wire a delay board into the existing power supply. pros: fairly easy cons: possibly requires an additional transformer due to the current power tranny being of unknown VA. Keeps the existing mediocre power supply. Some questions around best implementation.
Option b) is remove or bypass the existing supply with a valve rectifier and some regulation. Something like this example:  pros: Tube soft start, tube rectifier sound, upgraded b+ supply. Potentional all in one power supply solution if using the example board.
cons: extra cost upfront, extra upkeep (buy and store more tubes), increased complexity of packaging (tubes, extra transformer).
Option c) is remove or bypass the existing b+ supply with a proper regulated mosfet b+. Something like this  pros: proper regulated b+ supply. Doesn't require extra power tranny.
Cons: Questions around efficacy of soft start.
Option d) Modify and retrofit the existing board with some regulation and soft start. Pros: lowest cost, likely the smallest package. cons: Hardest option: difficulty of design, difficulty of fabrication, possibly much fabrication. Much head scratching.
|
Post by Owen Y on Oct 27, 2018 15:01:03 GMT 12
Over on diyaudio.com some people have tried moving the volume pot to the input side but it hasn't been successful Vol attenuator on the input of the line stage is most common & no problem usually. We are often tempted to solve one problem & can introduce new ones
|
Post by sadface on Oct 27, 2018 15:58:51 GMT 12
Over on diyaudio.com some people have tried moving the volume pot to the input side but it hasn't been successful Vol attenuator on the input of the line stage is most common & no problem usually. We are often tempted to solve one problem & can introduce new ones Hi Owen,
That is the exact situation I recall.
Moving the volume pot to the input side is easy. What proved more troublesome was figuring out what to put in place of the moved 50k pot.
It might be worth me having another dig through the relevant threads to see if anybody ended up with a good implementation.
I recall thinking the whole operation looked like far more trouble that it was worth to fix something that seems to not be broken.
|
Post by colinf on Oct 27, 2018 19:13:54 GMT 12
The mosfet regulator schematic you mentioned shows no mosfets, only bipolars, and an LM 317 regulator. IME anything with bipolar regulation you can hear. They make the sound flatter. Mosfets work better for B+. You could always add a choke to the B+ supply which would smooth it better than a 2k resistor. The other way to get delayed B+ would be to use a HT switch that switches on the B+ separately. That’s a cheap solution but you need to operate it manually. You’d wait for the valve heaters to warm up for 15sec before switching it on. To get lowest distortion out of the 3875, the input impedance needs to be the same as the negative feedback (nfb) impedance. (As it has a differential bipolar input stage.) In the LM3875 data sheet there is a 1k resistor in series with the input that is there to match the 1k nfb resistor in the schematic. On your valve pre the output impedance would be in the order of 1k making having the volume pot in that position superfluous, as it increases the output impedance beyond 1k at anything other than full volume or zero volume. It would be better at the input of the circuit. That’s the theory for lowest distortion. But not so fast! You mentioned you like the sound of the tube pre with the line source connected straight to the input with no pot and that the pot on the output sounds nicer. I can understand that as it would make the sound tubey, which is what you want. So to get sonically unpleasant distortion out of the 3875 you could insert a jfet input opamp (like an opa134) set to a gain of 1 (a buffer circuit) placed inbetween the valve preamp and the 3875, and keep the 1k resistor at its input there, and leave the volume pot at the output of the valve buffer. The distortion of an opa134 is about two orders of magnitude less than the LM3875. With the high input impedance of the jfet opamp the volume control impedance can be quite high if necessary so you can use the switched volume control as normal.
AMR-iFi R&D
|
Post by sadface on Oct 27, 2018 21:31:09 GMT 12
Hi Colin,
Thanks for the insight there. I have always thought the 50k was an important part of the circuit and was where it was for a reason. One thing that just struck me is that right now I am treating this as 2 separate amplifiers.
Moving the volume control to the input side always made sense from a that's how it's normally done point of view. But if I view it as 2 cap coupled stages within a single amplifier then it really doesn't make sense to leave the volume control where it is.
Does this look about right?
I thought heavily about putting the B+ on its own power switch but decided that might be a bit of a pain for the missus. It needs to just work.
Thanks for the pointing out the bipolars. Would something like this be more suitable?
|
Post by colinf on Oct 28, 2018 2:45:14 GMT 12
That would work. On your LM3876 circuit diagram the 220uf and 22uf caps at the bottom of the diagram are marked as the wrong polarity. The caps will explode if you build it like the circuit diagram says. The 10k resistor on the input of the valve pre (R1) will reduce the volume at low volume control settings with a 50k pot. The gain is about 55 with that setup which is a lot of gain with a sensitive speaker. The 22k resistor at the input of the LM3876 can be changed to 47k to achieve the 50k or so output loading on the valve preamp. The other regulator you’ve shown is also a bipolar transistor design. I was thinking of this for B+ soft startup: 
AMR-iFi R&D
|
Post by sadface on Oct 28, 2018 16:12:21 GMT 12
Hi Colin,
They acknowledge the issue with that circuit diagram in the notes for the build. They corrected it on the actual pcbs.
So I had a good go over the board today; much squinting and turning the board over and over. I can confirm the signal stage is correct. The power supply was largely correct but with a few inaccuracies, one glaring.
Here is an updated psu schematic.
Thanks for that design, looks interesting.
I was having a bit of a ponder on actual parts for your schematic this morning. Here is what I came up with. I didn't get around to figuring out a suitable zener as I'm rather unfamiliar with them at this stage. I'm guessing to correct the psu now would be this?
|
Post by colinf on Oct 28, 2018 22:43:31 GMT 12
Yes, but it might be better to connect the negative of the heater supply to 0V instead of positive. You don’t need a 1k resistor there, ground it directly. That way the heater voltage will be positive relative to ground. You could even float the heater circuit a few tens of volts above ground which would reduce the induction of hum from the heater circuit into the valve cathodes. The IRFBC20PBF mosfet would work nicely. How did you find out about it? You don’t need the 1N4007 diode from source to drain as the mosfet already has one built in. BTW any zener diode of voltage rating 10 to 15 volts will do. It protects the mosfet gate to source voltage from going above 20v (according to the data sheet) which would destroy it. It also discharges the 1uf capacitor when the power is turned off. I usually use 1w or 400mW ones. In operation there will be around 3v between the gate and source. As the mosfet drain sits at high voltage and the TO220 transistor metal tab is the drain, the mosfet should be insulated from the heatsink well so that you don’t get zapped if you accidentally touch it while it’s on.
AMR-iFi R&D
|
Post by sadface on Oct 30, 2018 16:20:15 GMT 12
Hi Colin,
I would have guessed on connecting the negative heater rail instead if the positive rail but I was working off what I saw in front of me. Any good reason why they would have gone off the positive rail at the factory?
I was lucky on the IRFBC20BF, I had a look for 600v MOSFETs on RS components. It looked likely and was also the cheapest one I could order 1 of.
Regarding zener diodes. This should suffice then?
I've got some TO220 insulation kits lying around somewhere I'm pretty sure.
Regarding floating the heater circuit a might above ground. Looking at some of the fancier psus I've seen. I'm guessing this is done by grounding the 0v rail from the heater supply to the junction of a voltage divider. Like so?
|
)
