[BUILD] Hairball Audio/MNATs JE-990

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Echo North

Well-known member
Joined
Jan 21, 2006
Messages
4,452
Location
Seattle, WA
One of the key features of the Lola Microphone Preamplifier (http://hairballaudio.com/lola/) is the inclusion of two 990 op-amps.  Deane Jensen’s now famous JE-990 paper describing the design, testing, and theory behind a powerful ultra low noise operation amplifier, was first introduced in the  February 1980 AES journal. The design started with the LM394 dual super matched transistor as the input device.  The LM394 is actually a monolithic device consisting of two pairs of 50 transistors connected in parallel to eliminate random variations in the individual devices.  In addition to being designed for low noise, the JE-990 is a powerful op-amp capable of adequately driving a 75Ω load.

Many commercial versions of the 990 design are available for purchase today.  The John Hardy Co JH-990 (http://www.johnhardyco.com/990OpAmpDetails.html) is certainly the best and most well respected version.  Luckily, it’s also one of the most affordable.  An equally authentic version is Steve Hogan's (http://www.soundsteward.com/) TSS-990-C which is unpotted but not in 2520 format.  Most of the other designs available for purchase fall short of the original design and are “990″ op-amps in name only.  Shortly after the release of the Lola, I began considering offering a DIY JE-990 that would be affordable, authentic, and easy to build.  Or at least as easy as soldering over 30 components on a one inch square PCB can be.  Certainly it would need to a 2520 footprint and I would need to source an affordable LM394 substitute.  Luckily much of this work was completed a few years ago by MNATs when he designed a DIY JE-990 2520 layout (mnats.net/JE-990.html).  Many here have built this version and much info can be found in the original support thread (http://www.groupdiy.com/index.php?topic=14422.0).  I contacted MNATs about reworking the PCB for a different input component to replace the discontinued LM394 and he agreed to give a shot.

The common replacement for the LM394 is the Analog Devices MAT12. The MAT12 is the successor to the MAT02 and is available in a TO-78 thru-hole package.  A very elegant  component, but not a cheap one.  They can cost $15-$30 depending on quantity.  Luckily there is another option.  Analog devices also offers the SSM2212 which is the surface mount version of the MAT12 (same die) and offers even better performance specs.  Even better, it can be purchased for a fraction of the cost of the LM394.

FULL_Groupdiy2.jpg


Pictured is the result.  You can see the SM2212 is mounted on the bottom.  What you may not see are the two 0.1uF SMD power supply bypass capacitors at the supply pins.  In true MNATs fashion the layout is, on its own, a work of art.  CR3 is a diode connected transistor matching Q3 and thermally coupled using a piece of heat shrink.  The output transistors are thermally coupled to biasing diodes using the clip described in the JE-990 paper.  The thermally coupled biasing diodes provide enough current limiting stability for the op-amp to drive a shorted load indefinitely with no heat sink.  All of the components used are those quoted in the JE-990 paper or their modern equivalent.

The DIY-990 arrives with the bottom layer SMD parts (SSM2212 and two bypass capacitors) already soldered to the PCB.  The six PCB pins are also pre-soldered.  All that remains are 14 resistors, 11 diodes, 8 transistors, 3 capacitors, and 2 inductors.  To make it easier an assembly guide is available for download (http://www.Hairballaudio.com/images/990/DIY-990-Manual.pdf).

DIY_Groupdiy.jpg


Information on the fully assembled JE-990 and DIY-990 kit can be found at:
http://www.Hairballaudio.com/990/

The PCBs are made and assembled in the USA.
 
1. Would these 990 OpAmps work in Classic API stuff like their 312DI mic pres or the LC53 EQs? In other words is it a direct replacement for 2520 style op amps?
2. Do they run at 16 and 24V mode? (for the 51X rack modules with 24V capability)
 
rob61 said:
1. Would these 990 OpAmps work in Classic API stuff like their 312DI mic pres or the LC53 EQs? In other words is it a direct replacement for 2520 style op amps?
In general, yes they should drop in.


rob61 said:
2. Do they run at 16 and 24V mode? (for the 51X rack modules with 24V capability)

The assembled version ships with a 15/16V option or a 24V option.  The only difference is two resistor values (R10,R11).  The DIY version ships with 33K resistors for R10,R11 (15/16V) however, if you exchange these two resistors for 62K resistors you'll be set up for 24V supplies.  This is covered on the assembly guide resistor map page.

Thanks,

Mike
 
Thanks for the quick reply. Just ordered a pair to try out in some of my modules. What is the result if I change out the resistors to accomodate 24v but then use them in a 16V module? Do I have to switch them back again?
 
Awesome!
Looks great.
Good idea with the clamps to thermally couple the diodes.
Any audio experience yet? Does this new version sound good next to a JH990c or DIY w/ mat12?
 
dmp said:
Awesome!
Looks great.
Good idea with the clamps to thermally couple the diodes.
Any audio experience yet? Does this new version sound good next to a JH990c or DIY w/ mat12?

I think they sound great!  The SMD SSM2212 is the mat12 in a different package.  It's the same part just way cheaper.  In terms of performance they are very close to the JH-990-C.  DIY 990 has 0.5dBu more EIN than the Hardy, which at 130dBu is really only measurable.  The slew rate is 18V/1.1uS rather than 18V/1uS.  Again only measurable.  A big part of this is that John custom trims a component for balance and obviously uses different components, layout, and has decades of experience with this circuit.  Everything else measured spot on.

The clamps are the same part specified by D. Jensen in the paper.

Mike
 
rob61 said:
Thanks for the quick reply. Just ordered a pair to try out in some of my modules. What is the result if I change out the resistors to accomodate 24v but then use them in a 16V module? Do I have to switch them back again?

Really poor performance I would assume.  Would be power starved.

You should switch out the resistors.
 
dmp said:
The clamps are the same part specified by D. Jensen in the paper.

Waldom Clamp # C-201?
I'd like to add these to mine. Do you know where I could get some?

Yep.

OnlineComponents had them but I cleaned them out.  If you just need a few send me an email/PM.  You pay shipping.

Mike
 
Built a pair up last night. Very nice kit.
I noticed a small typo R12 100 ohms should be 110 ohm on page 9.
 
dmp said:
You sub'd the 62pf/91pf with 68pf/100pf?

Yes.  The subs are much easier to find.  I think modern day standard values have changed a little.  Saw no difference in testing vs Hardy's.
 
Yes.  The subs are much easier to find.  I think modern day standard values have changed a little.  Saw no difference in testing vs Hardy's.
In the paper by Deane Jensen he describes that C2,C3,R8 network as setting the open loop response compensation. The technical details go way over my head. A good sub if they work as well as the original 990. I'm hoping to do a side-by-side comparison soon.
FWIW, you can still get 91pf COG from Newark. Otherwise they are smd (I built 4 990s a few years ago with the 62pf in smd, which worked pretty well).
The preamp design at the end of the paper also calls for a 91pf for the feedback loop.
 
dmp said:
Yes.  The subs are much easier to find.  I think modern day standard values have changed a little.  Saw no difference in testing vs Hardy's.
In the paper by Deane Jensen he describes that C2,C3,R8 network as setting the open loop response compensation. The technical details go way over my head. A good sub if they work as well as the original 990. I'm hoping to do a side-by-side comparison soon.
FWIW, you can still get 91pf COG from Newark. Otherwise they are smd (I built 4 990s a few years ago with the 62pf in smd, which worked pretty well).
The preamp design at the end of the paper also calls for a 91pf for the feedback loop.

91pF is often quoted in pre-amp designs.  I use a 100pF in the Lola.    68pF and 100pF are definitely the modern conventional standard values. I'll be interested to hear if find a difference in your testing.  Let me know if you do, I'll revisit it.

I did a ton of testing vs the Hardy.  Only VERY small differences in slew and EIN which I attribute to the fact that John trims R4 and R5 for DC balance.

Keep me posted!

Mike
 
I tested this out side-by-side against a previous 990 I built (w/ lm394h) in the standard servo preamp design by D Jensen.
I couldn't tell the difference between them and they both sounded awesome. Way to go mike!

But of the two I built, one has really low output for some reason. A problem somewhere...
 
dmp said:
I tested this out side-by-side against a previous 990 I built (w/ lm394h) in the standard servo preamp design by D Jensen.
I couldn't tell the difference between them and they both sounded awesome. Way to go mike!

But of the two I built, one has really low output for some reason. A problem somewhere...

Awesome! (well not the one with problem)

Building a real 990 is not too hard.  If you follow the Jensen doc and have a good layout (thanks to MNATS!), you should end up with killer 990.

Mike
 

Latest posts

Back
Top