Tech Flashback: Even More Internal PCI Dial-Up Modems

Regular retro-technology service continues with this blog post. I showed some of my vast collection of dial-up equipment in the past, especially my internal PCI modems, but because of the bleak future of voice-band modem technology, I sought to obtain a few more different sorts of modems to add to my collection. The reason for this will become clear over the next few days when … the posting happens.

Anyhow, lets just delight in the sights of some more PCI modems.

Netcomm InModem IN5699_5

To be honest, I really didn’t need to get any more of these, but since I saw them at auction in their pristine sealed condition and I found one of my IN5699_4’s based on the same Agere 1648C chipset had somehow failed, I thought it was worth the money just to grab them so as to preserve them for the future.

2016030111312580 2016030111312581 2016030111312584 2016030111312585 2016030111312582 2016030111312583

This unit was towards the end of the reign of dial-up, and was one of their more “premium” internal modems, with a golden backplate to offset it. This was because it was one of two internal modems that Netcomm were selling, with a inferior IN5920 based on the Agere Soft-Modem DSP-less chipset rather than the Lucent/Agere WinModem 1648C “Mars” chipset which did have the DSP and could operate quite reliably even on rather slow machines. The LT Winmodems also had frequent driver updates, and good connect rates in my experience, whereas the DAA set-up on the inferior IN5920 was often significantly slower (52k vs 44k connects at my house) and so I ended up giving my IN5920 away.


They also went to a red PCB for this model, and it has the latest 1648C chipset on it, along with an Agere silicon solid state DAA, but still with a traditional hybrid transformer. Also unlike some of the last models to be produced, it had a very tinny piezo buzzer to provide audible feedback of the modem’s line condition, which really gave a poor rendition of the actual signals on the line especially when inside a computer case.


Serial number on the back, but of course, no warranty on these anymore since Netcomm is now more a company focusing on wireless solutions.


For the extra touch of class, the brand is stamped into the backplate as well, whereas the older IN5699_4’s had just an adhesive label.

2016030111332587 2016030111322586

You get a phone cable, driver and software CD that had Cheyenne Bitware for fax/voice/data operation (a very antiquated piece of software) and a quick start guide. Because it was a very common modem, and drivers were made available, it is plug-and-play even in Windows 7 x64.

Amigo AMI-2019F aka XH1136 Intel Host-Accelerated Modem (HaM)

This particular modem was sold quite a lot at Dick Smith, and the name “host accelerated” really sounded a lot better than “winmodem” or “soft modem”, and so at one stage I even wanted to get one myself.


Now that I have one, it doesn’t seem so exciting anymore, since even though it is Intel branded, Intel has no support for it anymore. And it’s not really Intel as such, because the chipset is an Ambient Technologies chipset (which Intel acquired). The modem has a conventional transformer, and a relay controlled line switch. It also has voice ports for speaker and microphone on the rear, and an inbuilt buzzer.


The PCB was dated Week 30 of 2001, and is a little dusty as it was a pull from someone’s machine.


The modem is not actually supported under Windows 7 at all from what I could find, and does not automatically install under XP. Instead, finding the driver is necessary, and as Amigo no longer exist, we have to thank the Internet Archive because they still have it.

Mitsubishi Diamond Voice 56P

This is another Agere/Lucent WinModem based device, and I managed to get this sort of by a happy accident. Ultimately, it has an HV90 chipset, which turns out to be an older 1646 DSP as opposed to the 1648C above.


It has a very compact design with a relay line switch.


Again, it is a little dusty, dated week 52 of 2001. Why this modem was of importance will be explained in a future posting.

PTI-401-H00 Motorola SM56


A later soft modem with an entirely silicon DAA, and quite a weak speaker. PCB is dated Week 14 of 2006, and I believe PTI stands for Puretek Incorporated.


The modem does have driver support and automatically installs under Windows 7 but has connectivity difficulties. Under XP it seemed to work just fine.

Agere Pinball P40


Because of an auction deal, I managed to receive a bucketload (10) of these units, which are extremely late units. It utilizes the Agere SV92PP chipset which is the nasty purely-soft chipset which the IN5920 would have used (or its earlier version). This also has a purely silicon DAA with a “digital” transformer, which is quite interesting.


The rear label states a manufacturing date of 11th October 2010. The card is plastered with approvals, because it appears to be either a pull or supplied for OEM pre-installation into their computers.


As a very cheap OEM product, they did everything to save money, which meant only fitting one jack to the modem for the line, and providing no looped-out phone port, given that many people didn’t use the phone near their computer and often they had no good reason to use it because the modem wasn’t switching the port or using it correctly in voice mode anyway. It was also plug-and-play under Windows 7 x64.

Broadcom BCM94212


This particular acquisition was from the USA, because these modems were never very popular in Australia at all. This was one of maybe two dial-up modems that Broadcom ever made, with their focus shifting to broadband and wireless technologies, and thus is significant in some way because of it. Despite this, it seems we may have been duped all along, since the chipset has the letters AGR in the printing, which suggests it has something to do with Agere.


Again, it is a pull, so it is a little dusty, but it seems to be an older OEM product, so it’s not entirely skimping on all the features and still has a phone port on the rear. It is dated Week 14 of 2003.


Why do I say it’s likely to be an Agere? Well, when it was tested, the unit certainly behaved very similarly to the Agere soft modem (SV92PP) tested above, and it sounded very much identical. That being said, the driver support for this was non-existent past Windows XP, unlike the Agere.


When I was looking at their driver .sys files, I spotted some similarities in the build naming, where AGRSM stood for Agere soft modem, and BCMSM stood for Broadcom soft modem. The way the releases are named as \build\Win32\opt*wdm*\bin seemed a little strange to be coincidental.


Some of the other strings were identical as well, and it seems the driver has (internally) some ability to identify the type of answering modem as well in both example files.


This highly suggests the DSP is definitely an Agere chipset, with the differences lying in the Agere driver probably only working with Agere DAAs, whereas the Broadcom didn’t opt to use Agere DAAs and instead substituted Silicon Labs Si3012/3021 instead, hence needing the custom driver. This also implies that the modem could have been possible to support under modern OSes if the vendor bothered upgrading the driver.

Speedlink 56PCI Retail


This was another modem I ordered from overseas, but got lost in the post on the first shipment causing these series of postings to be held up. This was a Smartlink SL2800 chipset, which apparently had something to do with ST’s chipsets, and eventually was bought out by Conexant (formerly Rockwell). These weren’t common at all, but the driver was in other ways, in products called the Modio, and modems based on CNR/AMR riser cards, as their soft-modem drivers were pretty much the Smartlink drivers.


This unit was sold as a Speedlink 56PCI Retail, and is a late model dated Week 29 of 2005.


A very generic backplate and make indeed.


To grow a collection of voice-band dial-up modems in 2016 may seem pointless and somewhat strange, and I would agree with you on that. But all of this was not done without something very specific in mind, to preserve something which might soon well be forgotten despite great interest in the recent months. More to come … stay tuned.

About lui_gough

I'm a bit of a nut for electronics, computing, photography, radio, satellite and other technical hobbies. Click for more about me!
This entry was posted in Computing, Telecommunications and tagged , , , . Bookmark the permalink.

2 Responses to Tech Flashback: Even More Internal PCI Dial-Up Modems

  1. rasz_pl says:

    Im sure you are aware of Oona Räisänen ( ), she did the now famous ‘The sound of the dialup, pictured’ writeup and graphic.
    I think at one point she was playing with implementing soft modem(or fax modem) entirely in perl (lol wtf)

    Just few weeks ago I went on a discovery hike trying to figure out how modems looked in the early days of home computing (eighties). I was expecting big pcb with rows of opamps and carefully tuned filters (bell 103 fsk). Instead I found that in 1982 Commodore released 1600 vicmodem starting a trend of cheap connectivity and services (quantum link, compuserve, BBSes), and at that time modems were already one chip integrated affair 🙁

    diy bell 103 modem's%20Circuit%20Cellar%2C%20Volume%203&pg=PA178#v=onepage&q=Ciarcia's%20Circuit%20Cellar%2C%20Volume%203

    Im still tempted to try and implement modem on a microcontroller, $2 72mhz stm32 should be enough for 33.6

    • lui_gough says:

      Well, since early modulations were the domain of FSK, they could be implemented rather easily with oscillators and multiplexers or even by “abusing” GPIO pins to generate the signal, and counting zero-crossing transitions on the receive. That wouldn’t be highly tolerant of noise, but it’s a similar strategy to that of old cassette tape recording (e.g. Kansas city standard IIRC). Of course, being nice to the phone company means band-limiting your signals, so you might chuck in an LPF, but low-bitrates could have been mostly achievable in a semi-“soft” way.

      – Gough

Error: Comment is Missing!